👋 ICYMI

The last piece of the puzzle landed Friday morning. It didn't make the picture any clearer. The August CPI report showed headline inflation at 3.4% annually and 0.4% monthly — both matching expectations. On the surface, nothing alarming. But underneath, the core reading — stripping out food and energy — came in at 0.3% monthly, above the 0.2% consensus and up from July's 0.2%. That hotter-than-expected core number is the one the Fed cares about most. And it landed four days before the most important rate decision of the year.

"The Fed is looking for confirmation that the disinflationary trend is back on track, and this report shouldn't give them that confidence," said Collin Martin, Schwab's head of fixed income research.

Markets ended the week lower overall but rallied on Friday as oil prices dipped, lifting stocks despite the mixed CPI data. The S&P 500 fell roughly 0.5% for the week — its second consecutive losing week — but bounced on Friday's final session. The Nasdaq and Dow also declined during the week.

Rate hike odds sit at 59% for the September 15–16 FOMC meeting, according to CME FedWatch. "If you have a CPI reading that surprises to the upside, that's going to really make it difficult for them not to hike rates," warned Mark Hackett, Nationwide's chief market strategist, earlier in the week. The core CPI didn't blow out — but it didn't cool either. Warsh now has the data he needs to make his case either way.

Consumer inflation expectations surged. The University of Michigan's one-year inflation outlook jumped to 4.6% from 4.0% — the highest since June. Longer-run expectations edged up to 3.4%. "With a resurgence in fuel prices and trade tensions, consumers anticipate greater pressures on their pocketbooks to come," said survey director Joanne Hsu. Consumer sentiment is now 16% below where it stood in February, before the Iran war began.

The 10-year Treasury yield hit 4.818% midweek — its highest level since November 2023 — before easing slightly on Friday. U.K. and Eurozone yields also rose to multi-year highs. The global bond market continues to price higher-for-longer.

Consumer discretionary stocks are down 6.5% since mid-August. Energy is the only S&P 500 sector with gains over the last month. Sector breadth is deteriorating — a sliver of the index that mostly reflects rising oil pulling money out of consumers' pockets.

🔁 Market Movers

• 📊 August CPI: Headline In Line, Core Hotter Than Expected

  The headline CPI at 3.4% annual and 0.4% monthly matched forecasts. But core CPI's 0.3% monthly gain — above the 0.2% consensus — suggests underlying price pressures haven't fully cooled. The acceleration from July's 0.2% core reading is the detail that matters most for the Fed. Shelter costs remain elevated. Services inflation is sticky. And the 0.4% monthly headline — up from 0.1% in July — shows that energy prices are re-accelerating through the consumer economy.

• 📉 Second Consecutive Losing Week — RSI Approaching Oversold

  The S&P 500 declined roughly 0.5% for the week — its second straight weekly loss — amid relentless pressure from rising yields and oil. The index's RSI dropped to 43, down from 65 just a month ago and approaching the 30 threshold that signals oversold conditions. Consumer discretionary led losses, down 6.5% since mid-August. Energy was the only sector positive over the past month. Friday's rally, driven by falling oil, prevented a worse weekly loss.

• 🛢 Oil Remains the Inflation Wildcard

  Oil prices dipped on Friday, helping stocks rally despite the CPI data. But WTI remains near $85 and Brent near $90 — well above the sub-$70 levels reached in late June when the peace deal was fresh. The resumed U.S. strikes on Iran, combined with Strait of Hormuz disruptions that continue intermittently, keep energy costs elevated. The August CPI's 0.4% monthly increase was driven primarily by energy — and until oil sustainably breaks lower, headline inflation won't fall meaningfully below 3%.

• 📈 12-Month Inflation Expectations Jump to 4.6%

  The University of Michigan's one-year inflation expectations surged from 4.0% to 4.6% — the highest reading since June and a sharp reversal from the gradual cooling trend that began in spring. Longer-run expectations rose to 3.4%. Rising inflation expectations are one of the Fed's biggest concerns — when consumers expect higher prices, they tend to demand higher wages, creating a self-reinforcing cycle that's harder to break.

👀 Signals I'm Watching

• 🏛 The Fed Meets Tuesday — Hike, Hold, or Surprise?

  Kevin Warsh chairs his third FOMC meeting on September 15–16 with rate hike odds at 59%. The case for a hike: 162,000 August jobs, 3.4% headline CPI, 0.3% core CPI (above consensus), and 4.6% one-year inflation expectations. The case for a hold: consumer sentiment near historic lows, consumer discretionary down 6.5%, the RSI approaching oversold, and oil-driven inflation that could fade if the Iran situation stabilises. Edward Jones expects this to be a "live meeting" where both outcomes are genuinely possible. The dot plot and Warsh's press conference will matter as much as the decision itself.

• 📊 The Core CPI Problem Is Real

  The headline CPI matching estimates gave the market temporary relief. But the core reading — which the Fed watches most closely — accelerated from 0.2% to 0.3% monthly. That's the wrong direction. If September's core CPI (due October 10) confirms the upward trend, the "transitory energy-driven inflation" narrative that has sustained equity valuations all year will be under severe threat. The Fed can tolerate headline inflation driven by oil because it's outside their control. Core inflation driven by services and shelter is inside their control — and demands a response.

• 📉 The Market Is Approaching Oversold — Which Can Be Bullish

  The S&P 500's RSI at 43 — down from 65 a month ago — is approaching the 30 level that typically signals oversold conditions and precedes bounces. If the Fed holds on Tuesday, the relief rally could be sharp and swift, as positioning is deeply defensive. If the Fed hikes, the initial selloff could take the RSI below 30 — creating a technical buying opportunity for those with a longer time horizon

• 🏭 The Two-Speed Economy Is Getting Worse

  Energy is up 43% year to date. Consumer discretionary is down 2.3%. Nike $NKE ( ▼ 2.24% ) is at a 20-year low. Travel stocks are at 52-week lows. But the S&P 500 is positive for the year and on track for its fourth consecutive annual gain. This divergence between the index (propped up by AI and energy) and the consumer economy (stressed by inflation and high borrowing costs) is the defining tension of 2026. The Fed's decision Tuesday will either ease that tension (hold) or intensify it (hike).

$ODD over the past five days

⚡️ Sharing my exact portfolio holdings and trades in real time is one of the core benefits of my Investment Club membership. This past week, my main contrarian position, Oddity Tech $ODD ( ▼ 4.62% ) , surged over 30% after posting earnings results that were well ahead of analyst expectations.

Members knew about my position, average purchase price, and portfolio weighting — and were positioned accordingly.

The next earnings season is just around the corner. If you'd like to follow along, get full access to my deep dives, and join a community of long-term growth investors where I share my exact portfolio, now is a good time to become a member.

Thank you for your continued support,

George

⚠️ Red Flag to Note

The Market Isn't Pricing a Hike — It's Pricing Uncertainty

Rate hike odds at 59% mean the market is essentially a coin flip on the most important monetary policy decision of the year. That level of uncertainty, by itself, is the problem. When the market can't decide whether the Fed will tighten or hold, it underprices both outcomes — and the reaction to whichever one lands is amplified. A hold at these odds could spark a 2–3% relief rally. A hike could send the S&P 500 toward 7,400 or lower, with the selloff concentrated in the rate-sensitive sectors that have already been under pressure — consumer discretionary, housing, and small caps. The VIX's relative calm doesn't reflect the magnitude of what's at stake. Tuesday isn't just another Fed meeting. It's a 59/41 bet on the direction of markets for the rest of 2026.

🔍 Insider Transactions I’m Watching

📬 Closing Note

The data this week didn't make Kevin Warsh’s job easier. Headline CPI at 3.4% was in line. Core at 0.3% was hotter than expected. Inflation expectations jumped to 4.6%. The labour market added 162,000 jobs last month. And consumer sentiment is 16% below where it stood before the war.

Both sides of the debate have ammunition. The hawks have sticky core inflation, a resilient labour market, and rising expectations. The doves have deteriorating consumer spending, approaching oversold technicals, and the argument that oil-driven inflation is outside the Fed's control. The market is at 59/41. A coin flip with trillion-dollar consequences.

But here's what I keep coming back to: regardless of whether Warsh hikes or holds, the underlying forces that have powered this market all year haven't changed. AI demand is accelerating. Corporate earnings are growing 25%+. And the broadening of the rally means the bull case isn't dependent on a single stock or a single Fed meeting.

Stay patient. Stay selective. And let the data guide the story.

Until next Sunday —

George ☕

The Physical AI trade has become one of the market’s hottest themes, but you wouldn’t know it from looking at Ouster $OUST ( ▼ 1.26% ) . After an extraordinary rally earlier this year, the stock has fallen sharply from its highs and recently returned to levels last seen in August. Yet the underlying business continues moving in the opposite direction.

$OUST performance in recent months

Customers are deploying more Ouster sensors, shipments are reaching records, revenue growth is accelerating and several partnerships suggest its technology is finding its way into everything from autonomous robots and agricultural equipment to drones and intelligent traffic systems.

Ouster provides something increasingly important to the Physical AI ecosystem: the ability for machines to understand the physical world around them. And as autonomous machines move from demonstrations into real-world deployment, demand for that capability could increase dramatically.

The recent correction may have created another attractive entry point into one of the more interesting picks-and-shovels plays on Physical AI. Let’s take a closer look.

$ODD surging post-earnings today

As a contrarian investor, I'm always looking for non-obvious opportunities that most investors are ignoring. Oddity Tech $ODD was one of them. The company reported better-than-expected earnings today, sending the stock up as much as 30%.

Investment Club members had access to my portfolio and were positioned accordingly.

If you'd like full access to all our deep dives, including my price target for each stock, and want to join our Investment Club — where I share my exact holdings — you can become a member here.

George

👋 ICYMI

Two weeks ago, the economy lost 23,000 jobs and the stock market rallied to a record high. This week, the economy added 162,000 jobs and the stock market fell. Welcome back to "good news is bad news."

Friday's August nonfarm payrolls report showed the U.S. economy added 162,000 jobs — more than three times the 53,000 consensus — while unemployment held steady at 4.1%. June and July payrolls were both revised upward, effectively erasing last month's shocking job losses. The labour market isn't weakening — it's resilient. And that's exactly what the market didn't want to hear.

The S&P 500 fell 0.38% on Friday to 7,718.60. The Dow dropped 272 points to 53,414. The Nasdaq dipped 0.29% to 26,507. The 2-year Treasury yield surged to its highest level since January 2025. And Fed funds futures immediately repriced a 58% probability of a rate hike at the September 16–17 FOMC meeting — up from 35% just 24 hours earlier.

The week was volatile beyond the payrolls report. Tuesday opened with losses after the U.S. launched fresh strikes on Iranian targets, sending the S&P 500 down 0.71% and the Nasdaq down 1.03% — the worst opening day of September in years. Wednesday brought a recovery as Treasury yields briefly paused their ascent. Thursday saw a strong bounce as the Nasdaq and S&P 500 leaped on a weaker dollar. Then Friday's payrolls report reversed it all.

The 10-year Treasury yield hit 4.818% on Wednesday — its highest since November 2023 — before pulling back slightly on Friday. U.K., German, and French yields also rose to multi-year highs. The global bond market is sending a unified message: rates aren't coming down anytime soon.

Nike $NKE ( ▼ 2.24% ) hit a 20-year low of $38.07 on Tuesday — the lowest price for the stock since 2006. Travel and entertainment names including Wynn Resorts $WYNN ( ▼ 1.2% ) , Las Vegas Sands $LVS ( ▼ 2.67% ) , VICI Properties $VICI ( ▼ 1.12% ) , and Carnival $CCL ( ▼ 2.0% )  all touched 52-week lows the same day. While AI and chip stocks dominate headlines, entire sectors of the consumer economy are quietly deteriorating.

Meanwhile, Bill Gates made one of the largest insider purchases of the year. Cascade Investment, Gates' investment vehicle, bought approximately $305 million in Republic Services $RSG ( ▼ 1.24% ) shares on September 2 — bringing his cumulative stake to $388.8 million across 35 purchases concentrated in May and August.

🔁 Market Movers

• 💼 August Payrolls: 162,000 vs 53,000 Expected — Rate Hike Revived

  The August jobs report demolished the "cooling labour market" narrative. Employers added 162,000 jobs — triple the consensus — and both June and July were revised upward. The three-month average is now running well above the sub-50,000 pace that had convinced markets the hike was dead. Fed funds futures immediately repriced to a 58% probability of a September rate hike, up from 35% on Thursday. "Good news is bad news — and this was very good news," said one Wall Street strategist.

• 📈 10-Year Yield Hits 4.818% — Highest Since November 2023

  The benchmark 10-year Treasury yield reached its highest level in nearly three years midweek before settling slightly lower after Friday's selloff. The 2-year yield — the most sensitive to Fed policy expectations — jumped to its highest since January 2025 on Friday. Global bond markets moved in tandem, with U.K. gilts and Eurozone yields also rising. The yield curve's message is unambiguous: the bond market expects higher rates.

• 🛢 U.S. Strikes Iran Again — September Opens in the Red

  The first trading day of September began with U.S. strikes on Iranian targets, pushing oil higher and sending all three major indices into the red. The S&P 500 fell 0.71% and the Nasdaq dropped over 1% on Tuesday. Despite the June peace deal, military operations have resumed intermittently throughout the summer. The market has largely priced through the strikes, but each escalation adds to the background yield pressure that has kept the 10-year above 4.7%.

• 👟 Nike at 20-Year Low — Consumer Weakness Deepens

  Nike hit $38.07 on Tuesday — its lowest price since 2006 — as the consumer discretionary sector continued to deteriorate. Energy is up 43% year to date. Consumer discretionary is down 2.3%. The divergence reflects an economy where corporate profits are strong but the consumer is under stress from elevated gas prices, high borrowing costs, and persistent inflation.

• 💰 Insider Selling Overwhelms Buying on September 2

  September 2 saw $175 million in insider selling versus just $2.6 million in purchases — one of the most lopsided sell-to-buy ratios of the year. Caterpillar's $CAT ( ▼ 0.06% ) CEO sold $26.2 million. Synopsys' $SNPS ( ▼ 3.6% ) Executive Chair sold $21.4 million. Airbnb's $ABNB ( ▼ 1.37% ) CSO sold $13.5 million. Abercrombie $ANF ( ▼ 3.88% ) insiders dumped $6 million. The broad-based executive selling at elevated valuations is a pattern that typically precedes periods of consolidation or correction.

👀 Signals I'm Watching

• 🏛 The Fed Meets September 16–17 — The Most Important Meeting of the Year

  Kevin Warsh chairs his second FOMC meeting in ten days with rate hike odds at 58%. The August payrolls report gives hawks the ammunition they need: a labour market that's adding jobs, not losing them. But oil has fallen from $112 to $85 since March, core CPI is trending toward 2.5%, and consumer sentiment remains near historic lows. The case for a hike exists — but so does the case for patience. Warsh's decision will define the trajectory of markets through year-end.

• 📉 Insider Selling at $175M vs $2.6M in Buys — Corporate Executives Are Cashing Out

  When Caterpillar's CEO, Synopsys' chair, and Airbnb's CSO all sell on the same day — totalling over $60 million among just three executives — it's worth paying attention. Insider selling alone doesn't predict corrections. But when sell-to-buy ratios reach 67-to-1 in a single session at the S&P 500's second-highest close of the year, it signals that the people running these companies see better value in cash than in their own stock at current prices.

• 📈 Energy +43% YTD, Consumer Discretionary -2.3% — The Two-Speed Economy

  The S&P 500 is on track for its fourth consecutive annual gain. But beneath the index, two economies are playing out simultaneously. Energy companies — boosted by the Iran war, elevated oil, and structural demand — are having their best year since 2022. Consumer-facing businesses — Nike at 20-year lows, travel at 52-week lows, retail sales declining — are telling a story of stress. This divergence can persist as long as AI-driven earnings growth offsets consumer weakness. But if Q3 earnings reveal cracks in the consumer sectors, the index's resilience will be tested.

$BE surges on S&P500 inclusion

⚡️ As my Investment Club members know, Bloom Energy ($BE) — the disruptive energy company — is my biggest portfolio holding. After surging 20% during the week, the stock extended its gains after market close on news of its inclusion in the S&P 500.

In a deep dive published on April 1, I shared my thesis on $BE, explaining why the stock was a strong buy despite its already solid performance. Shares have nearly doubled since then — and Investment Club members were positioned accordingly.

If you'd like to follow along, get full access to my deep dives, and join a community of long-term growth investors, you can become a member here.

Thank you for your continued support,

George

⚠️ Red Flag to Note

The September Rate Hike Would Be the First in Three Years

If the Fed hikes on September 17, it would be the first rate increase since July 2023 — a dramatic reversal of the easing cycle that markets had been pricing since late 2024. At the start of 2026, the consensus was for three rate cuts this year. Instead, the market is now pricing a hike. The journey from "three cuts" to "one hike" has been driven entirely by the Iran war's impact on oil, the resulting inflation spike, and now a labour market that refuses to cool convincingly. A September hike would push the fed funds rate to 3.75–4.00%, tightening financial conditions for an economy that's already showing consumer stress. Historically, surprise hikes late in a cycle — when the market had been expecting easing — produce the sharpest equity corrections because positioning is wrong-footed. The next ten days will determine whether Warsh follows through or buys more time.

🔍 Insider Transactions I’m Watching

📬 Closing Note

Two weeks ago, the economy lost jobs and the market hit a record. This week, the economy added 162,000 jobs and the market fell. If that doesn't capture the absurdity of the current moment, nothing does.

The market isn't confused — it's trapped. It needs a labour market weak enough to prevent rate hikes but strong enough to sustain earnings growth. August's payrolls report delivered the opposite of what it wanted: proof that the economy is resilient, which is great for Main Street but terrible for a Wall Street that has been banking on the Fed staying on the sidelines.

The next two weeks will shape the rest of the year. The Fed on September 17. The August CPI report on September 10.

After six months of war, peace deals, record IPOs, and the biggest AI earnings season in history, it comes down to one question: will Kevin Warsh hike?

Stay patient. Stay selective. And let the data guide the story.

Until next Sunday —

George ☕

Hims & Hers $HIMS ( ▼ 3.33% ) is still frequently viewed as a telehealth company selling treatments for hair loss, sexual health and, more recently, weight loss. But that description increasingly misses the bigger picture.

The company is building a broader consumer healthcare platform capable of launching new specialties through the same technology, brand, provider network, fulfillment infrastructure and customer acquisition engine. Each successful category potentially makes the next one easier to scale.

The latest quarter provided further evidence that this strategy is working. Growth accelerated, subscriber momentum improved and newer healthcare categories continued gaining traction even as the company invested heavily in its international expansion and broader product portfolio.

Weight loss remains important, but the long-term opportunity is becoming much larger than GLP-1s. The investment thesis ultimately comes down to whether Hims can become one of the dominant consumer healthcare platforms globally. The evidence increasingly suggests that it can. Let’s see why.

$HIMS has been volatile this year

Is $SEZL a buy following the recent selloff? Discussed in last week’s deep dive

One of my newest positions is Sezzle ($SEZL) — a BNPL platform that is rapidly evolving into a comprehensive consumer finance company. In last week's deep dive, I shared my thesis, price target, and portfolio allocation with members.

If you'd like full access to all our deep dives, including my price target for each stock, and want to join our Investment Club, you can become a member here.

George

👋 ICYMI

After six months of newsletters covering wars, ceasefires, peace deals, record earnings, and record gold, it all came down to one company, one quarter, one Wednesday.

On Wednesday evening, Nvidia $NVDA ( ▲ 0.57% ) answered. Revenue of $96.22 billion — up 106% year over year and 4.5% above estimates. Adjusted EPS of $2.22 versus the $2.09 consensus. Data centre revenue of $89 billion, up 117% and now comprising 92% of total sales. And Q3 guidance of $108 billion — $3.8 billion above what Wall Street expected.

On Thursday, Nvidia surged 8.7% — its biggest single-day gain since April 2025 — adding $442 billion in market value in a single session. The S&P 500 rose 0.7% to 7,731 and the Nasdaq gained 1.6%. But the rally was remarkably narrow: technology was the only positive sector on Thursday, with the other ten S&P 500 sectors finishing in the red.

The week delivered another blockbuster. Salesforce $CRM ( ▼ 1.46% ) surged 22.6% on Thursday after reporting EPS of $5.90 versus $3.27 expected — a near-80% beat — on $11.35 billion in revenue. It was a powerful counter to the narrative that AI software companies can't monetise — Salesforce just proved they can.

On Friday, PayPal $PYPL ( ▼ 0.41% ) plunged 16% after Bloomberg reported that Advent International and Stripe had abandoned their $53 billion takeover pursuit. The collapse of what would have been the largest fintech deal in history leaves PayPal without a strategic bid and resets the stock to pre-deal-rumour levels.

Gold continued its extraordinary August run, with UOB forecasting that the metal is on track for its strongest monthly gain since September 1999 — up roughly 15% in August alone. Spot gold hit $4,720, a three-month high, supported by Treasury buybacks, a weakening dollar, and persistent hedging demand.

And at Jackson Hole, Fed Chair Kevin Warsh delivered his first major policy speech on Friday. Markets awaited clarity on whether the September 16–17 FOMC meeting would bring a hike, a hold, or a surprise.

🔁 Market Movers

• 🤖 Nvidia: $96.2 Billion. $108 Billion Guided. Questions Answered.

  Revenue more than doubled year over year to $96.22 billion. Data center revenue hit $89 billion (+117% YoY), now 92% of total sales. Adjusted gross margins held at 74%. Q3 guidance of $108 billion ±2% exceeded the $104.2 billion consensus by 3.6%. Amazon announced a deal to buy 2 million Nvidia GPUs and deploy Nvidia's Vera CPU. CFO Colette Kress said top-five hyperscaler capex is expected to reach $1.3 trillion next year, up from $800 billion in 2026. Supply commitments more than doubled to $279 billion. Nvidia returned $26 billion to shareholders via buybacks and dividends during the quarter. And the company disclosed $7.8 billion in gains on equity investments — including stakes in Intel and SpaceX.

  CEO Jensen Huang's message was clear: "The shift to agentic AI requires 15 to 100 times more compute." Dan Ives of Wedbush told CNBC the AI trade is "only in the third inning".

• 💻 Salesforce +22.6% — AI Software Can Monetize After All

  Salesforce's Q2 blowout — EPS of $5.90 versus $3.27 expected, revenue of $11.35 billion — was the strongest earnings reaction in the software sector all year. After ServiceNow's 18% crash in April, IBM's worst day in 40 years, and months of underperformance in the software complex, Salesforce showed that AI-powered enterprise software can deliver revenue acceleration and margin expansion simultaneously. The stock's 22.6% jump was its best day in over two years.

• 💳 PayPal -16% — $53 Billion Deal Collapses

  Stripe and Advent International walked away from their takeover pursuit of PayPal, sending shares down 16% on Friday. The deal, which would have valued PayPal at roughly $53 billion, was first reported in late July and had provided a floor under the stock. Without a strategic bid, PayPal faces renewed pressure. Recall that PayPal's CFO bought $255K in shares on June 15 — the day the Iran peace deal was announced — and the stock is now below that purchase price.

• 📊 Warsh at Jackson Hole — The September Signal

  Fed Chair Kevin Warsh delivered his first Jackson Hole speech on Friday. Markets parsed every sentence for clues about whether the September 16–17 FOMC meeting would bring a rate hike. With the VIX at 14.51 and the S&P 500 near highs, the market is pricing in a hold — but Warsh's hawkish reputation means any hint of tightening bias could quickly reverse the risk-on sentiment.

👀 Signals I'm Watching

• 📈 Hyperscaler Capex to $1.3 Trillion in 2027

  Nvidia's disclosure that top-five hyperscaler capital expenditure is expected to reach $1.3 trillion next year — up from $800 billion in 2026 — is the most important number in the earnings report. It means the AI infrastructure buildout is accelerating, not plateauing. When Microsoft $MSFT ( ▼ 1.64% ) , Amazon $AMZN ( ▼ 2.02% ) , Alphabet $GOOG ( ▼ 1.24% ) , Meta $META ( ▲ 0.7% ) , and Oracle $ORCL ( ▼ 3.07% ) together plan to spend $1.3 trillion in a single year, it validates every semiconductor, memory, networking, and power company in the supply chain. The concern that the AI capex cycle was peaking has been definitively answered — at least for the next 12–18 months.

• 🤖 The AI Software Narrative Just Changed

  Salesforce's 22.6% surge is as important as Nvidia's 8.7%. For months, the market's verdict on AI was clear: hardware earners win, software spenders lose. Salesforce just proved that AI-powered software can deliver 80% earnings beats, margin expansion, and revenue acceleration simultaneously.

• 🥇Gold's Best Month in 27 Years Is Saying Something

  When gold gains 15% in a single month — its best since 1999 — alongside record equity highs, it's not a contradiction. It's a signal. Gold is pricing in dollar weakness, fiscal deterioration ($1.8 trillion deficit), and the possibility that the Treasury's bond buyback intervention is the beginning of a longer period of financial repression. If the gold move is right, the dollar's decline will eventually feed back into import prices, complicating the Fed's inflation fight just as CPI appeared to be fading.

• 📊 The Narrowness Problem

  Nvidia's +8.7% and Salesforce's +22.6% powered Thursday's market gains. But ten of eleven S&P 500 sectors closed in the red the same day. The rally is real, but it's narrow — increasingly dependent on a handful of AI-linked names to carry the index. That's sustainable when those names deliver (as they did this week), but it creates fragility when they don't.

$CRWD surge on Thursday post earnings

🦅Shares of CrowdStrike ($CRWD) — the cybersecurity leader and one of my favorite holdings — surged as much as 20% on Thursday after the company posted its strongest quarter ever.

Investment Club members were positioned accordingly. I first opened a position in CrowdStrike back in July, taking advantage of a period of short-term weakness despite the widespread view that the stock was overvalued. The Club knew the full thesis from day one.

Pumpkin spice latte season feels like the right moment to get back on financial track — which is why I've put together a limited-time offer for newsletter readers. If you'd like to follow along, get full access to my deep dives, and join a community of long-term growth investors, you can become a member here.

Thank you for your continued support,

George

⚠️ Red Flag to Note

The Third Inning Means Seven More to Go

Dan Ives says the AI trade is "only in the third inning." Nvidia's supply commitments doubled to $279 billion. Hyperscaler capex reaches $1.3 trillion next year. These are extraordinary numbers — and they assume a future that requires continuous execution over the next five to seven years.

In the third inning, the game's direction is clear but far from decided. The risk isn't that AI demand slows tomorrow — it's that the market has already priced five years of growth into today's valuations. Nvidia trades at roughly 35x forward earnings. Salesforce jumped 22% in a day. The SOX index is up over 60% year to date. If at any point the revenue growth rate decelerates — not declines, just decelerates — the repricing could be swift. The AI trade has proven its fundamentals in 2026. The question for 2027 and beyond is whether execution can match expectations that are now measured in trillions.

🔍 Insider Transactions I’m Watching

📬 Closing Note

Six months ago, we started covering the first week of the Iran war. Oil was surging past $100. The S&P 500 was entering correction territory. Consumer sentiment was collapsing. And the question dominating every market conversation was whether the world was heading for a global recession.

In the months since, we've covered the war's escalation, the ceasefire, the peace deal, the Fed's most divided meeting in 34 years, the SpaceX IPO, Intel's resurrection, Micron's trillion-dollar moment, Apple's CEO transition, Meta's cloud launch, the Treasury's bond market intervention, gold's best month since 1999, and now — Nvidia's $96 billion quarter.

Through all of it, the underlying story has been remarkably consistent: AI demand is real, it's accelerating, and it's generating the kind of corporate earnings growth that absorbs almost any macro shock.

For investors who have been patient, selective, and guided by the data — this has been one of the most rewarding stretches in market history. The S&P 500 is up roughly 10% year to date. Corporate earnings are growing 25%+. And the AI trade has broadened from one company into an entire investment ecosystem.

It's been an extraordinary six months. Thank you for reading every Sunday.

Stay patient. Stay selective. And let the data guide the story.

Until next Sunday —

George ☕

Sezzle $SEZL ( ▼ 2.06% ) delivered another strong quarter earlier this month, yet the stock sold off sharply after earnings. At first glance, that reaction looks strange. The company beat expectations, raised guidance again, continued expanding its subscriber base and maintained strong profitability. But beneath the headline results, investors focused on a few softer signals: higher spending, lower sequential earnings and an outlook that implies slower growth during the second half. Those concerns are not completely unreasonable after the stock’s enormous run earlier this year. But the selloff may have gone too far.

$SEZL stock’s summer selloff

The underlying business continues to scale, credit performance remains healthy, subscriber engagement is improving and management is deliberately investing more aggressively in customer acquisition while returns remain attractive. At the same time, newer products are beginning to push Sezzle beyond traditional buy-now-pay-later and toward a broader consumer-finance platform. The recent correction has therefore created a much more interesting setup. Let’s take a closer look at what investors may be missing.

$LIME over the past 30 days

You may have missed it, but Lime ($LIME) — the scooter company — has become one of my favorite stocks lately. In last week's deep dive, I broke down my full thesis on this under-the-radar name and shared my price target for the year. The stock is up nearly 10% since then and has rallied nearly 50% over the past month — at a time when AI stocks have moved in the opposite direction.

If you'd like full access to all our deep dives and want to join our Investment Club, you can become a member here.

George

👋 ICYMI

This was a week the bond market dominated — and the government blinked first. The S&P 500 fell 1.4% to close at 7,674 on Friday — its first weekly decline since late July — as rising Treasury yields dragged equities lower through most of the week. The Nasdaq dropped 2% to 26,180. The Dow slipped to 53,277. AI and semiconductor stocks continued to sell off from their highs, extending the rotation that began in late July.

The catalyst for the selling wasn't earnings or geopolitics — it was yields. Treasury yields resumed their march higher midweek, with the 10-year climbing back to 4.74%, threatening the 5% level on the 30-year that has served as a psychological ceiling for equities all year. AI stocks sank — the Nasdaq posted three consecutive sessions of losses through Tuesday, falling 1.3% on Tuesday alone as chip stocks led the decline.

Then the Treasury stepped in. The government announced it would double its bond buyback programme for 10- to 30-year debt — an extraordinary move designed to inject liquidity and stymie the yield surge. The announcement initially pushed yields and the dollar lower, sparking Friday's rebound. The Dow surged 518 points (+0.98%) on the news. The S&P 500 gained 0.43%. But it was a band-aid, not a cure — the 10-year still closed the week at 4.74%, well above levels that historically constrain equity valuations.

While equities struggled, alternative assets surged. Gold hit a three-month high of $4,680 on Friday — its fifth consecutive weekly gain and its longest winning streak since October 2025. Gold is up roughly 6% on the week. And Bitcoin posted a staggering 22% weekly gain — its biggest move in months — driving Robinhood $HOOD ( ▼ 3.39% ) up 14% and Coinbase $COIN ( ▼ 10.1% ) up 8% on Friday

All eyes now turn to Wednesday: Nvidia $NVDA ( ▲ 0.57% ) reports Q2 earnings on August 26. After the $500 billion capital mobilisation announcement, the SpaceX $SPCX ( ▼ 3.15% ) chip deal, and a stock trading near $220, this is the most anticipated single earnings report of the year. If Nvidia delivers, the AI trade could reignite. If it doesn't, the selloff in semis could accelerate into something more painful.

🔁 Market Movers

• 🏛 Treasury Doubles Bond Buybacks — Government Intervenes in Yield Surge

  The Treasury announced it will double its buyback programme for 10- to 30-year government debt — a rare intervention designed to ease liquidity pressure and prevent yields from spiralling higher. The move initially pushed yields and the dollar lower, lifting gold and sparking Friday's equity rebound. But the 10-year still closed at 4.74%, and strategists questioned whether buybacks alone can sustainably cap yields while the federal deficit approaches $1.8 trillion and the Fed holds rates at 3.50–3.75%.

• 🥇 Gold Hits 3-Month High — Fifth Straight Weekly Gain

  Gold climbed to $4,680 — its highest level since mid-May and its longest weekly winning streak since October 2025. The rally is being driven by a weaker dollar on the buyback news, persistent yield uncertainty, and institutional hedging ahead of Nvidia earnings. Gold rising while equities fall is a classic late-cycle signal — it suggests investors are buying insurance, not just growth.

• ₿ Bitcoin Surges 22% — Biggest Weekly Move in Months

  Bitcoin posted a 22% weekly gain, its strongest performance since early 2025. Crypto-related stocks rallied sharply on Friday — Robinhood jumped 14% and Coinbase gained 8%. The move appears driven by the Treasury's intervention weakening the dollar, renewed institutional interest in crypto as a yield alternative, and technical momentum. Whether this is the start of a sustained leg higher or a short squeeze remains to be seen.

• 📉 First Weekly S&P 500 Decline Since Late July

  The S&P 500's 1.4% weekly loss broke its winning momentum, with AI and chip stocks leading the decline. Friday's rebound (+0.43%) on the Treasury buyback news prevented worse damage, but the index finished below 7,700 for the first time in two weeks. Healthcare (Merck $MRK ( ▼ 0.15% ) , J&J $JNJ ( ▲ 0.33% ) ) and financials (Goldman Sachs $GS ( ▼ 1.19% ) ) led Friday's recovery, reinforcing the rotation from tech to defensive/cyclical sectors.

• 🛢 Oil at $87 — Still the Inflation Wildcard

  Crude oil remained near $87 on WTI and $89 on Brent, elevated by ongoing Middle East tensions. The June CPI progress (3.5% → 3.4%) that gave the market confidence is at risk of reversing if oil stays at these levels through the autumn. The 10-year yield's persistent strength at 4.74% partly reflects the bond market's view that energy-driven inflation hasn't been fully resolved.

👀 Signals I'm Watching

• 🤖 Nvidia Wednesday: The Most Important Earnings Report of the Year

  Nvidia reports Q2 earnings on August 26. Analysts expect roughly $90 billion in revenue. The $500 billion capital mobilisation, the SpaceX exclusive chip deal, and the $678 billion in hyperscaler commitments all point to a beat. But the stock is trading at 35x+ forward earnings, the semiconductor sector is selling off, and the bar has never been higher. After Microsoft $MSFT ( ▼ 1.64% ) added $450 billion in a day and Palantir $PLTR ( ▼ 0.43% ) surged 29%, the market has shown it will reward genuine AI earners. If Nvidia delivers, it could reverse the chip selloff and push the S&P 500 back toward 7,800+. If it disappoints, the correction accelerates.

• 💰 Gold + Bitcoin Surging Together = Hedging Is Real

  When gold and Bitcoin both rally sharply in the same week, it typically signals that investors are hedging against something — currency debasement, yield instability, or equity fragility. Gold's 5% gain and Bitcoin's 22% surge alongside a declining equity market is the kind of cross-asset signal that appeared before the 2020 and 2022 corrections. It doesn't guarantee a selloff, but it tells you that smart money is buying insurance.

• 📊 The Treasury's Buyback Has Limits

  Doubling bond buybacks is a liquidity injection, not a fundamental solution. The deficit is approaching $1.8 trillion. The Fed is holding rates. And the 30-year yield nearly hit 5.2% two weeks ago. If the buybacks fail to contain yields — or if the market interprets the intervention as a sign of desperation rather than confidence — it could backfire. The next 30-year Treasury auction will be the test of whether the government's move has genuinely eased conditions or merely delayed the reckoning.

• 🔄 Russell 2000 at 3,000 — Small Caps Are Having Their Moment

  The Russell 2000 crossed 3,000 this week — a milestone for small-cap stocks that have been the quiet winners of the rotation out of mega-cap tech. The index is up 22% in the first half and continues to benefit from falling rate hike expectations and improving domestic economic conditions. If the AI trade narrows further, small caps could become the market's new leadership.

$LIME performance over the past month

🧐 The best opportunities are often hiding in the most unexpected places.

In my most recent deep dive, I covered Lime ($LIME) — a name most investors have overlooked entirely. My newly opened position rose over 10% shortly after, on top of a 50% rally over the past month. As I shared in the deep dive, $LIME may still be significantly undervalued, and members have full access to my price target for the year.

If you'd like to be first to know about opportunities like this and join a community of growth investors, you can become a member here.

George

⚠️ Red Flag to Note

Three Assets Are Telling Three Different Stories

Equities are saying the economy is fine — the S&P 500 is 1.8% from its all-time high. Gold is saying investors are nervous — five straight weekly gains to a three-month high. And Bitcoin is saying the dollar is weakening — a 22% weekly surge, its strongest in months.

These three signals rarely point in different directions for long. When they diverge like this, one of them is usually wrong. Either equities are right and gold/Bitcoin are overreacting to yield noise, or gold/Bitcoin are right and equities haven't yet priced in the risks that the alternative assets are screaming about. Nvidia's report on Wednesday could be the catalyst that resolves the divergence — either confirming the AI-powered bull case or giving gold bugs the validation they've been waiting for.

🔍 Insider Transactions I’m Watching

📬 Closing Note

This was the week the government had to step in. The Treasury doubling its bond buyback programme isn't a routine policy adjustment — it's an admission that the yield market was moving in a direction that threatened financial stability. When the 30-year yield pushed toward 5.2% two weeks ago and the 10-year settled at 4.74% this week, the pressure on equities, mortgages, and corporate borrowing costs became too much to ignore.

Whether the intervention works remains to be seen. But what's already clear is that the market's relationship with yields has become the defining dynamic of the second half. The first half was about AI earnings and geopolitics. The second half is about whether the bond market allows equity valuations to hold.

Gold knows. Five straight weeks of gains. Three-month highs. Investors are hedging. Bitcoin knows. A 22% weekly surge says the dollar is losing a fight that most equity investors haven't fully noticed.

But the stock market hasn't made up its mind yet. It's sitting near all-time high, waiting for one event to tip the balance. That event arrives Wednesday. Nvidia's Q2 earnings will either confirm that AI demand is strong enough to overpower the yield headwinds — or it will give the bond market permission to impose its will on equity valuations.

Stay patient. Stay selective. And let the data guide the story.

Until next Sunday —

George ☕

The micromobility industry does not exactly have a great reputation among investors. The sector burned billions of dollars during its first growth wave, competition was brutal, scooters were treated as disposable assets, and several once-promising companies ultimately collapsed. Bird, once valued at more than $2 billion, became the poster child for everything that went wrong.

That history may explain why the recent IPO of Neutron Holdings $LIME ( ▼ 0.78% ) , the company behind Lime’s electric bikes and scooters, received relatively little attention. But Lime today looks surprisingly different from the micromobility companies investors remember.

The company has spent years improving vehicle durability, building its own hardware and software stack, increasing utilization and establishing scale that smaller competitors struggled to achieve. More importantly, its first earnings report as a public company provided further evidence that the model is working. ⁠

Lime might therefore be one of the more interesting recent IPOs because expectations surrounding the company remain relatively low. That also probably explains the stock’s strong early performance. Let’s take a closer look.

$LIME has rallied since the IPO in early July

AI spending is entering a new phase. On August 10, Nvidia $NVDA ( ▲ 0.57% )  announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish compute financing platforms designed to mobilize more than $500 billion of third-party capital over time. The announcement matters for more than Nvidia.

It suggests that AI infrastructure is becoming a finance and credit-market story, not just a corporate spending story. GPUs, data centers, networking systems, power equipment, cooling infrastructure, and leases may increasingly be evaluated like long-duration infrastructure assets. That could accelerate the buildout. It could also introduce a new layer of risk.

📈 The Nvidia AI funding push, explained

The headline number is large. The structure is more important. Nvidia says the new platforms will help finance AI factories across its ecosystem, including frontier AI labs, enterprises, and AI cloud providers. The arrangements are currently based on memorandums of understanding, meaning the final agreements, terms, and timing remain subject to execution.

The $500 billion target is therefore not a committed fund. It is better understood as a potential pool of financing capacity that could be deployed across multiple projects over time.

The goal is straightforward: allow customers to build or access AI compute without funding every GPU cluster and data center entirely from their own balance sheets. That changes the investment equation.

Previously, investors primarily watched hyperscaler capital expenditure announcements. Now, they also need to watch who owns the infrastructure, who lends against it, how leases are structured, and whether the underlying compute generates enough cash flow to support the financing.

🔍 Why compute is becoming an asset class

Nvidia’s core argument is that modern compute has characteristics investors typically associate with infrastructure:

• Scarcity: Demand for high-end AI compute remains ahead of available capacity in many markets.

• Revenue generation: GPUs can support cloud services, model training, inference, and enterprise workloads.

• Transferability: Nvidia argues that compute can be used across different customers and workloads.

• Ecosystem support: CUDA software and Nvidia’s developer base may extend the useful life of its systems.

• Long-duration demand: AI adoption could create multi-year demand for computing capacity.

This is the central idea behind the Nvidia AI funding push. Instead of treating GPUs only as equipment purchased by a technology company, financial institutions may underwrite them as productive assets that generate recurring usage revenue.

The analogy is closer to a data center, power plant, or toll road than a conventional consumer electronics product.

That does not make the assets risk-free. It means Wall Street is beginning to develop financing structures around them.

What the financing could accelerate

The most obvious impact is on data-center construction. Many AI projects face a familiar constraint: demand exists, but the customer may not have enough capital, credit capacity, or patience to build the required infrastructure upfront. Third-party financing can help bridge that gap.

The capital may flow through several layers of the AI infrastructure stack. This is why AI infrastructure investing extends well beyond semiconductor manufacturers.

The best-positioned companies may be the ones solving physical bottlenecks that cannot be fixed with software alone. Power and cooling equipment, for example, are becoming essential as rack densities rise.

Our previous breakdown of the AI buildout and its physical bottlenecks looked at why megawatts, thermal management, and optical bandwidth could remain critical investment themes.

The winners may not be obvious

Nvidia is the most direct beneficiary if financing helps customers buy more of its systems.

The company could benefit in several ways:

1. Higher hardware demand as more customers gain access to capital.

2. Greater ecosystem lock-in as financed projects are built around Nvidia’s platform.

3. More software adoption through CUDA and related tools.

4. Stronger customer reach among AI companies that cannot fund large clusters independently.

5. Potential recurring demand as operators refresh and expand their infrastructure.

But the broader opportunity includes companies involved in the surrounding infrastructure. Networking suppliers may benefit because larger GPU clusters require faster communication between processors. Memory manufacturers may see demand rise as AI systems require more high-bandwidth memory. Cooling providers could gain as conventional air cooling becomes less practical for dense deployments.

Power is another major theme. AI data centers require reliable, high-volume electricity, and grid connections can take years to secure. Hyperscalers are increasingly exploring on-site generation, power purchase agreements, and alternative sources of firm power.

That makes the on-site power trend an important part of the AI data center stocks conversation. The opportunity is broad. The quality of each business is not.

⚠️ The risks behind the financing boom

Financing can accelerate growth. It can also amplify mistakes.

Leverage

If AI infrastructure is financed with significant debt, operators need predictable revenue and high utilization to meet interest and lease obligations.

A data center with unused capacity is not automatically a productive asset. It is a capital-intensive facility with ongoing power, maintenance, and financing costs.

Technology obsolescence

GPU technology evolves quickly. A system that is highly valuable today may become less competitive after a new architecture arrives. Nvidia’s argument is that software compatibility, transferability, and continued demand can support residual values, but lenders still need to assess how quickly the hardware depreciates.

This is especially important when debt maturities extend beyond the useful life of a particular generation of chips.

Utilization

The financing model works best when compute is consistently used. Investors should watch whether AI cloud providers can maintain high utilization rates, secure long-term customer contracts, and charge enough for compute to cover operating and financing costs. A slowdown in model training or weaker inference demand could pressure returns.

Leases and hidden commitments

Some infrastructure may be owned by financing vehicles and leased back to technology companies or AI customers.

That could reduce the appearance of traditional capex while creating long-term contractual obligations. Investors should read footnotes carefully and track lease liabilities, purchase commitments, and minimum capacity payments.

Reported capital expenditure may not tell the full story of the resources committed to AI.

Slower-than-expected AI returns

The largest risk is simple: AI revenue may take longer to develop than expected.

Hyperscalers can tolerate substantial investment if AI services eventually generate strong returns. Smaller companies and highly leveraged operators may have less room for delays.

If customer demand, pricing, or model economics disappoint, the pressure could spread from operators to equipment suppliers, lenders, and asset owners.

What hyperscaler capex 2026 may be hiding

The market has focused heavily on hyperscaler capex 2026. That remains an important indicator, but it may become less complete as financing structures evolve.

A company can increase its AI capacity through:

• Direct purchases of GPUs and data-center equipment.

• Long-term leases.

• Capacity contracts with AI cloud providers.

• Joint ventures with infrastructure investors.

• Special-purpose vehicles that own the physical assets.

• Project financing backed by future compute revenue.

The economic result may be similar: more AI capacity comes online, even if the accounting presentation differs. That is why I’m watching cash returns on infrastructure, not just the size of spending announcements. The key question is whether new capacity is tied to credible demand and durable pricing.

Nvidia’s official announcement on the compute financing platforms makes the strategic intent clear. The final investment case will depend on the details that have not yet been disclosed.

A practical framework for long-term investors

For investors evaluating AI infrastructure opportunities, I would use five checks.

1. Follow the bottleneck

Look for the part of the stack where supply is constrained and difficult to expand quickly.

That could be advanced packaging, high-bandwidth memory, optical connectivity, power equipment, or liquid cooling. Bottlenecks can support pricing power : but only while demand remains healthy.

2. Separate orders from revenue

Large backlogs are useful. They are not the same as recognized revenue or free cash flow.

Check whether a company is converting orders into shipments, revenue, operating profit, and cash. Delays in construction or customer deployment can stretch the cycle.

3. Examine customer quality

A long-term contract with a well-capitalized hyperscaler carries a different risk profile from a speculative commitment by a small AI cloud provider.

Customer concentration, payment terms, contract duration, and renewal conditions all matter.

4. Test the utilization assumptions

Ask how much capacity must be used for the project to break even.

A financing structure built on near-perfect utilization may look attractive in a bullish scenario but become fragile if demand softens.

5. Stress-test depreciation and rates

Model what happens if GPUs depreciate faster, interest rates remain higher, or compute pricing falls.

The strongest businesses should retain reasonable economics under less favorable assumptions. That resilience is more valuable than a headline growth rate.

The bottom line

Nvidia’s AI funding push could mark an important transition. AI infrastructure is moving from a corporate procurement decision toward a broader financial market. Institutional capital may help fund the GPUs, servers, buildings, networking, power, cooling, and leases required for the next phase of adoption. That is constructive for the ecosystem. It could accelerate data-center construction and widen access to compute.

But financing does not create demand by itself. The long-term winners will likely be the companies with durable bottlenecks, strong balance sheets, credible customers, and cash flows that remain healthy when utilization or pricing cools. The risks are worth monitoring ; particularly leverage, residual values, lease commitments, and the timing of AI returns.

I’m staying focused on the builders, but with tighter risk controls and more attention to the credit side of the cycle.

For readers following these themes, our Investment Club provides early access to deeper company research and ongoing discussion around the AI infrastructure market. You can also subscribe to The Latte for curated stock-market analysis delivered directly to your inbox.

Stay disciplined,

George

Disclaimer: This post is for informational purposes only and does not constitute financial advice. Always conduct your own stock market research or consult with a financial advisor before making any investment.

👋 ICYMI

This was the week that felt like a market holding its breath. The S&P 500 closed Friday at 7,785 — within striking distance of its all-time high set the previous Friday. The Dow slipped 0.6% to 53,732. The Nasdaq edged lower. No major breakout, no major breakdown. Just a market churning near the top, digesting two competing narratives that point in opposite directions.

The bullish narrative: inflation is cooling. Wednesday's July CPI report showed headline inflation at 3.4% annually — matching expectations and down from 3.5% in June. Core CPI came in at 2.5% — the lowest since early 2025 and approaching the Fed's 2% target. Money markets now price a less than 50% chance of a September rate hike, down from 64% just weeks ago. The inflation story that dominated markets from March through June — driven by oil and the Iran war — is fading.

The bearish narrative: the consumer is weakening. Friday's data showed July retail sales dropped by the most in over a year. The University of Michigan's preliminary August consumer sentiment index fell to 51 — below the 55 consensus and near the record lows set during the war. Combined with last week's negative payrolls print (-23,000 jobs), a picture is emerging of an economy where inflation is falling not because things are getting better, but because demand is slowing.

On the AI front, Nvidia $NVDA ( ▲ 0.57% ) dominated headlines. CEO Jensen Huang announced Monday that Nvidia has partnered with Apollo $APO ( ▼ 1.02% ) , Blackstone $BX ( ▼ 1.24% ) , BlackRock $BLK ( ▼ 0.65% ) , Brookfield $BAM ( ▼ 1.69% ) , Goldman Sachs $GS ( ▼ 1.19% ) , and KKR $KKR ( ▼ 2.22% ) to mobilize more than $500 billion in third-party capital for AI infrastructure. Nvidia also secured an exclusive chip supply deal with SpaceX $SPCX ( ▼ 3.15% ) , pushing shares past $220 and addressing concerns about customer concentration. Separately, Intel $INTC ( ▼ 0.05% )  upsized its common stock offering from $15 billion to $20 billion to fund its own AI buildout.

Memory stocks continued to outperform — SanDisk $SNDK ( ▼ 1.36% ) surged 14% on Thursday and another 7.4% on Friday. AMD $AMD ( ▲ 2.19% ) gained 6.5%. Meanwhile, Broadcom $AVGO ( ▼ 1.58% ) fell 5.9% and hyperscalers (Meta $META ( ▲ 0.7% ) , Oracle $ORCL ( ▼ 3.07% ) , Amazon $AMZN ( ▼ 2.02% ) ) all declined on the final session as the consumer data spooked growth investors.

🔁 Market Movers

• 📊 July CPI: 3.4% — Inflation Continues to Cool

  The Consumer Price Index rose just 0.1% in July, bringing the annual rate to 3.4% — down from 3.5% in June and matching consensus. Core CPI of 2.5% annually is the lowest in over a year and approaches the Fed's target zone. The report eased rate hike fears and pushed the S&P 500 near its record on Wednesday.

• 🤖 Nvidia Mobilizes $500 Billion for AI with Six Asset Managers

  Nvidia announced partnerships with Apollo, Blackstone, BlackRock, Brookfield, Goldman Sachs, and KKR to channel more than $500 billion in private capital toward AI infrastructure — data centres, energy systems, and networking. The move reframes Nvidia not just as a chip seller but as the orchestrator of AI's capital formation. Nvidia also secured an exclusive chip deal with SpaceX, diversifying its customer base beyond hyperscalers.

• 📉 Retail Sales Drop, Sentiment Falls — Consumer Stress Emerging

  Friday's data showed two clear warning signs. July retail sales fell by the most in over a year, suggesting consumers are pulling back. The University of Michigan's August sentiment index dropped to 51 — below the 55 consensus. Combined with July's negative payrolls print, the consumer side of the economy is cooling noticeably.

• 💾 Memory Stocks Surge — SanDisk +21% in Two Days

  SanDisk rallied 14% on Thursday and another 7.4% on Friday. Micron $MU ( ▲ 0.39% ) gained 2.3% on bullish analyst recommendations. AMD jumped 6.5%. The memory and compute trade continued to outperform even as broader tech struggled, reinforcing the theme that AI hardware demand remains robust while the application layer and consumer-facing tech comes under pressure.

• 🏗 Intel Upsizes Stock Offering to $20 Billion

  Intel increased its previously announced common stock offering from $15 billion to $20 billion, with proceeds earmarked for "general corporate purposes" — widely understood to mean AI chip manufacturing capacity. The offering dilutes existing shareholders but signals management's urgency to compete in the AI foundry race. Intel shares dipped slightly on the news.

👀 Signals I'm Watching

• 📈 Nvidia Earnings August 26 — The Quarter That Sets the Tone for H2

  Nvidia reports Q2 earnings on August 26. After Q1's $81.6 billion revenue beat and the $500 billion capital mobilization announcement this week, expectations are sky-high. Analysts expect roughly $90 billion in Q2 revenue. The SpaceX chip deal, the asset manager partnerships, and the continued AI infrastructure ramp all point to another beat. But the stock is trading at 35x+ forward earnings — any disappointment on margins, guidance, or demand visibility could trigger a selloff in a sector that's already shown signs of fatigue.

• ⚖️ Inflation Falling + Consumer Weakening = Goldilocks or Recession?

  The July CPI at 3.4% with core at 2.5% is exactly what the Fed wants to see. But it's falling partly because demand is softening — retail sales dropping, jobs contracting, and sentiment at 51. The bullish interpretation is Goldilocks: inflation cools enough to prevent hikes, while growth stays positive. The bearish interpretation is that the consumer is the leading edge of a slowdown that eventually hits corporate earnings. The next two months of data will resolve this debate.

• 🛢 Oil Creeping Higher Again — Watch the Strait

  WTI climbed to $84 midweek, its highest level of August. The Strait of Hormuz situation remains unresolved despite the June MoU. Any renewed escalation would reverse the inflation progress that allowed CPI to fall. The 10-year yield's rise to 4.73% suggests the bond market hasn't fully bought the "inflation is over" narrative.

• 💰 The U.S. Budget Deficit Hit $432 Billion in July

  The fiscal 2026 deficit is approaching $1.8 trillion. At some point, the bond market will price this more aggressively — particularly with the 30-year already above 5%. For now, equities are ignoring the fiscal picture entirely. That's typical in bull markets. It becomes a problem when it isn't.

One of the most popular holdings among Investment Club members; last week

💡 Nebius ($NBIS), one of the most popular holdings among Investment Club members, surged nearly 50% last week after blockbuster earnings results. We were positioned accordingly.

This week I'm publishing a deep dive on an under-the-radar stock with the potential for over 100% return in the medium term. It's a name most investors haven't heard of — and that's exactly why the opportunity exists.

Investment Club members will be the first to know. If you'd like to be in the room when it drops, you can join us here.

George

⚠️ Red Flag to Note

The Consumer Data Is Telling a Different Story Than the Stock Market

The S&P 500 is near its all-time high. Corporate earnings are growing 25%+. AI capex is accelerating. And yet payrolls went negative for the first time in three years. Retail sales just posted their biggest monthly drop in over a year. Consumer sentiment fell to 51 — below expectations and near the historic lows set during the Iran war. And the three-month payrolls average is running at roughly 35,000 jobs — barely above zero.

The stock market is a forward-looking machine, and right now it's looking at Nvidia's $500 billion capital partnerships and Microsoft's $678 billion backlog. The consumer data is looking at $4 gas, 3.4% inflation, and an economy that may be cooling faster than corporate earnings suggest. These two stories can coexist for a while. But not forever. If August and September payrolls confirm the weakening trend, Q3 earnings estimates — currently projecting 18% growth — will need to come down. And when estimates come down, multiples compress. Nvidia's August 26 earnings and the September 5 payrolls report are the two data points that will determine which story the market follows into autumn.

🔍 Insider Transactions I’m Watching

📬 Closing Note

This was a week of duelling narratives — and neither won decisively. On one side: inflation at 3.4%, core at 2.5%, rate hike odds below 50%, and Nvidia mobilizing $500 billion from the world's largest asset managers. The AI story has never been stronger. The infrastructure buildout has never been better capitalized. And the market is within a fraction of its all-time high.

On the other side: the economy lost jobs last month. Retail sales just dropped by the most in over a year. Consumer sentiment is at 51. And the University of Michigan survey suggests people feel worse about their finances than they did during the worst of the Iran war.

Both of these things are true at the same time. And they create a market that churns — one that can't break to new highs because the consumer data is weakening, but can't sell off meaningfully because AI demand and corporate earnings remain exceptional.

The resolution comes in the next three weeks. Nvidia reports August 26. Payrolls land September 5. And the Fed meets September 16–17. Those three events will determine whether this is a healthy pause before the next leg higher — or the beginning of a rotation that favours defence over offence.

For now, the data says stay invested, but tighten your focus. The market is rewarding selectivity over conviction, and precision over momentum.

Stay patient. Stay selective. And let the data guide the story.

Until next Sunday —

George ☕

AI is moving from the lab into everyday products. That changes the investment question. The next phase of AI spending may depend less on building larger models and more on running those models quickly, cheaply, and reliably for millions of users. This is where AI inference matters and where the AMD $AMD ( ▲ 2.19% )  versus Nvidia $NVDA ( ▲ 0.57% )  debate becomes more interesting.

For AMD vs Nvidia inference investing, Nvidia remains the stronger overall platform. AMD, however, offers a credible challenger with attractive hardware economics and more room to gain market share.

I’m watching both companies closely. The better choice depends on whether an investor prioritizes ecosystem strength and profitability or greater upside from a successful challenger.

Quick take: Nvidia looks better suited to investors seeking the stronger AI inference platform today. AMD may appeal more to investors willing to accept higher execution and valuation risk in exchange for potentially faster growth from a smaller base.

Training vs. inference: The simple explanation

AI training is the learning process. A company feeds a model enormous amounts of data, and the model adjusts its internal parameters to improve its answers. Training requires immense computing power and often involves thousands of GPUs working together. Inference is what happens after training.

When you ask a chatbot a question, generate an image, summarize a document, or use an AI coding assistant, the model is performing inference. It is producing an output based on what it has already learned.

The distinction matters for investors:

• Training: optimized for building and refining models.

• Inference: optimized for speed, latency, memory efficiency, and cost per response.

• Commercial scale: inference may eventually become the larger and more recurring workload as AI applications reach more users.

This is why the market for AI chips for inference could become a major long-term growth opportunity.

The two companies in one view

The comparison is not simply about which chip is faster. In production, the winning platform must combine hardware, software, networking, memory, support, and availability. That gives Nvidia an important advantage.

Inference performance: workload matters

Nvidia has the broader performance lead at the high end of the market. Its Blackwell platform is designed for demanding workloads, including large reasoning models and mixture-of-experts architectures. Nvidia reported that Blackwell systems running TensorRT software delivered more than three times the inference throughput of previous-generation Hopper systems on selected models and configurations. Those are vendor-reported benchmarks, so investors should treat them as directional rather than universal. Still, the signal is clear: Nvidia is optimizing the entire stack, not just the GPU. AMD’s advantage is more specific.

The Instinct MI300X includes 192GB of HBM memory and 5.325TB per second of memory bandwidth, compared with 80GB and 3.35TB per second for Nvidia’s H100 in the comparison published by AMD. More memory can allow certain large models to run with fewer GPUs, reducing communication overhead.

AMD’s own testing found that MI300X performed competitively with H100 on several large language model inference workloads. In some memory-bound scenarios : particularly models with long outputs or large context requirements : AMD reported a performance advantage. The important takeaway is not that AMD always beats Nvidia. It does not. Instead, AMD can be competitive when memory capacity, bandwidth, and cost per token matter more than absolute peak performance.

Software: Nvidia’s biggest moat

This is the most important section in any AMD AI stock analysis. Nvidia’s hardware is powerful. Its software ecosystem is arguably more valuable.

CUDA has been developed for nearly two decades and is deeply embedded across universities, startups, cloud providers, and large enterprises. The platform includes optimized libraries, compilers, debugging tools, and frameworks that help developers move from experimentation to production.

For inference, Nvidia adds TensorRT-LLM, Triton, NIM microservices, and other deployment tools. These products can reduce the time required to optimize and deploy models, which is valuable for customers under pressure to launch AI products quickly. Nvidia’s official CUDA platform shows how broad that ecosystem has become.

AMD is addressing the gap through ROCm. Its open-source approach supports frameworks such as PyTorch and vLLM, while HIP allows developers to adapt portions of CUDA code for AMD hardware.

ROCm is improving. AMD has also introduced the AMD Developer Cloud, giving developers access to Instinct GPUs without purchasing hardware.

But adoption is not only about technical capability. Developers value stability, documentation, third-party support, and day-one compatibility with new models. Nvidia still leads on those measures.

Pricing and gross margins

AMD’s strongest commercial argument is cost efficiency. If an AMD accelerator can deliver similar inference performance at a lower purchase price, cloud providers may have a reason to diversify their hardware base. Hyperscalers operate at enormous scale, so even a modest reduction in cost per token can become meaningful. The challenge is that hardware price is only one part of total cost.

Software migration, engineering time, support, power consumption, networking, and utilization all influence the economics of an inference deployment. Nvidia’s ecosystem can justify a premium if it reduces implementation friction and improves the percentage of time that hardware is productively used.

The financial results reflect this difference. Nvidia reported $130.5 billion of fiscal 2025 revenue and a 75% GAAP gross margin. AMD reported $34.6 billion of fiscal 2025 revenue, a 50% GAAP gross margin, and a 52% non-GAAP gross margin. AMD’s data-center segment generated $16.6 billion in revenue during the year.

Nvidia captures more profit from each dollar of AI infrastructure spending. AMD has more potential operating leverage if its AI business scales successfully.

Customer adoption and competitive positioning

Nvidia’s customer adoption remains its clearest advantage. Its GPUs are widely available through major cloud providers and are already integrated into enterprise AI workflows. That installed base reinforces the CUDA ecosystem, which in turn makes Nvidia hardware easier to purchase and deploy.

AMD is becoming a more credible second source. Microsoft and AMD have highlighted work involving DeepSeek inference on MI300 accelerators in Azure. AMD has also pointed to partnerships with Meta, Cohere, Microsoft, and Red Hat as evidence of broader ROCm adoption.

This matters because hyperscalers do not want to depend on a single supplier indefinitely. Custom silicon from Google, Amazon, Microsoft, and other large customers is another part of the competitive landscape.

The long-term market may not be winner-take-all. Nvidia can retain the premium workloads while AMD wins deployments where customers prioritize openness, supply diversity, or cost per token.

Valuation: the part investors should not skip

Nvidia’s valuation risk is straightforward. The company has delivered extraordinary growth and margins, but the market already recognizes its quality. If AI infrastructure spending slows, product transitions become difficult, or gross margins normalize faster than expected, the stock could face pressure even if the business continues to grow.

AMD carries a different valuation risk. Because AMD is viewed as a share-gain story, investors may already be pricing in substantial future AI revenue. If Instinct adoption grows more slowly than expected or ROCm fails to close the software gap, the stock’s growth premium could compress.

Rather than relying on a single price-to-earnings figure, I would compare:

• Data-center AI revenue growth

• Gross-margin direction

• Free-cash-flow conversion

• Large customer concentration

• Evidence of production inference deployments

• Valuation relative to expected earnings growth

The latest share prices and analyst estimates should always be checked before making an investment decision.

Risks worth monitoring

Nvidia

• Valuation expectations: exceptional growth may already be reflected in the share price.

• Ecosystem dependence: CUDA is a moat, but customers and developers are actively exploring alternatives.

• Custom silicon: hyperscaler-designed chips could reduce demand for merchant GPUs in selected workloads.

• Architecture changes: new model designs may reduce the importance of Nvidia’s current performance advantages.

AMD

• Execution risk: AMD must deliver reliable products, supply, and software improvements at the same time.

• ROCm adoption: technical progress does not automatically translate into developer preference.

• Customer concentration: a small number of hyperscalers could account for a large portion of AI accelerator demand.

• Margin pressure: competing on price may limit the profitability of AMD’s AI business.

• Changing workloads: an advantage in dense, memory-bound models may not carry over to every future architecture.

Which stock may suit which type of investor?

Nvidia may be the better fit for investors who value:

• A mature and widely adopted software ecosystem

• Strong customer visibility

• Higher gross margins

• Broad exposure to training, inference, networking, and enterprise AI

• A lower execution burden relative to AMD

AMD may be more appealing to investors who value:

• Potential market-share gains

• Cost-optimized inference

• Open-source software development

• High memory capacity for large models

• Greater upside if hyperscalers diversify their accelerator suppliers

This is not a recommendation to buy either stock. It is a framework for thinking about the trade-off.

AMD vs. Nvidia inference investing: the bottom line

Nvidia is still the stronger all-around AI inference company in 2026.

Its hardware roadmap, customer base, software ecosystem, and profitability create a powerful combination. For investors seeking the more established AI infrastructure leader, Nvidia appears to offer the clearer foundation.

AMD is the more execution-sensitive opportunity.

Its hardware can be competitive in important inference workloads, and its open software strategy could gain traction as customers focus more heavily on cost per token. But the investment case depends on AMD closing the software gap while expanding production adoption and protecting margins.

I’m keeping Nvidia in the “platform leader” category and AMD in the “credible challenger” category. The most important signals from here will be inference revenue growth, ROCm adoption, gross-margin trends, and whether hyperscalers move from testing AMD hardware to deploying it at meaningful scale.

For more research on AI infrastructure, growth stocks, and the companies building the next market cycle, explore The Latte’s stock picks or join the Premium community. That is where we follow the details beyond the headline : including early access for members of the Investment Club.

Stay diversified, keep risk controls tight, and watch how the economics of inference develop.

George

Disclaimer: This post is for informational purposes only and does not constitute financial advice. Always conduct your own stock market research or consult with a financial advisor before making any investment.

The AI infrastructure trade continues to evolve, but one thing has become increasingly clear — hyperscalers are not slowing down. After months of concerns that the enormous amounts of capital flowing into AI infrastructure could eventually overwhelm returns, recent results from the largest cloud companies have provided important evidence to the contrary. AI demand remains extremely strong, capacity remains constrained in parts of the market, and companies are increasingly showing that the billions being invested in infrastructure are translating into actual revenue.

Microsoft $MSFT ( ▼ 1.64% ) recently reported that its AI business surpassed a $37 billion annual revenue run rate, growing 123% year over year, while Azure revenue increased 40%. The company has also continued expanding physical AI capacity, including a new Texas data center campus expected to add roughly 2 GW of capacity.⁠

That matters enormously for the companies supplying the infrastructure underneath those AI clusters, and one of the most interesting beneficiaries remains Credo Technology $CRDO ( ▲ 0.2% ) .

$CRDO has nearly fully recovered from the recent selloff

Credo provides the high-speed connectivity technology that allows GPUs, servers, switches and other components inside modern data centers to communicate with each other. As AI clusters become larger, the amount of data moving between those components explodes, making networking increasingly important to overall system performance. And Credo has positioned itself directly in that bottleneck.

Following another exceptional year of growth, continued hyperscaler spending and an expanding product portfolio across both copper and optical connectivity, the long-term thesis remains bullish. Let’s take a deeper dive.

$NBIS, a top Investment Club holding

Back on March 11, I published a deep dive on Nebius ($NBIS) — one of my favorite AI cloud infrastructure names — explaining why the stock was due for a significant move. Fast forward to today and shares have rallied over 130%, including a 28% surge today alone.

Investment Club members were the first to know, with full access to the deep dive, my thesis, personal portfolio allocation, and price target.

If you want to be first to know about similar opportunities in the future, you can join us here.

George

👋 ICYMI

The U.S. economy lost 23,000 jobs in July. The stock market responded by rallying to a new all-time high. Welcome to the "bad news is good news" regime.

Friday's July payrolls report was a shock — nonfarm payrolls contracted by 23,000, far below the 83,000 consensus, the first job losses in over three years. Unemployment dipped to 4.1% from 4.2%, and labour force participation slipped to 61.4%. Rather than panicking, the market celebrated: the weak data convinced traders the Fed won't need to raise rates, effectively killing the September rate hike consensus that had weighed on equities for months.

The S&P 500 surged to a record close of 7,757.64 — up 3.6% on the week, its best weekly performance since April. The Nasdaq jumped 5.2% to 26,690 — its strongest week since May — with semiconductor stocks leading the charge, the SOXX ETF gaining over 5%. The Dow closed at 54,037.

Gold had its best week in seven months, surging 7.5% to $4,348 as the collapsing rate hike narrative drove investors into precious metals.

Monday had set the tone for the week. The Dow hit a new record close of 53,178 — up 693 points — as oil slid on renewed Iran peace talk reports and Treasury yields fell 6 basis points. The S&P 500 gained 1.48% and the Nasdaq rose 2.1%.

But the star of the week was Palantir $PLTR ( ▼ 0.43% ) . The company reported Q2 revenue of $1.94 billion — up 93% year over year — with U.S. commercial revenue climbing 149%. Full-year revenue guidance was raised to $8.15 billion. CEO Alex Karp called the results "otherworldly." The stock surged roughly 29% on Tuesday, approaching its best single-day gain ever.

🔁 Market Movers

• 💼 July Payrolls: -23,000 — First Job Losses in Over Three Years

  The economy shed 23,000 jobs in July, versus expectations for 83,000 added. It's the first negative payrolls print since early 2023. Unemployment dipped to 4.1% as the participation rate fell, suggesting some workers are leaving the labour force rather than being fired. Prior months were also revised lower. The data landed like a permission slip for the Fed to hold — and the market traded accordingly, with rate hike probabilities for September collapsing from 64% to below 20%.

• 📈 S&P 500 Hits Record 7,758 — Best Week Since April

  The S&P 500's 3.6% weekly gain — closing at an all-time high of 7,757.64 — was its strongest week in nearly four months. The rally was broad: semis led with the SOXX up 5%, but healthcare, industrials, and financials all participated. Salesforce $CRM ( ▼ 1.46% ) (+2.47%), Nvidia $NVDA ( ▲ 0.57% ) (+2.33%), and Honeywell $HON ( ▲ 1.02% ) (+2.28%) were the Dow's top performers on Friday. The Nasdaq's 5.2% gain was its best since May.

• 🤖 Palantir: "Otherworldly" — Revenue +93%, Stock +29%

  Palantir's Q2 report was the earnings event of the week. Revenue of $1.94 billion (+93% YoY), U.S. commercial revenue up 149%, GAAP operating margin expanding, and full-year guidance raised to $8.15 billion. CEO Karp said: "We have managed to build something wholly unique." The stock's 29% single-day surge on Tuesday sent its market cap past $350 billion and confirmed Palantir as the premier pure-play AI software platform. Unlike Meta's $META ( ▲ 0.7% ) free cash flow collapse, Palantir is growing and generating cash simultaneously.

• 🥇 Gold Surges 7.5% — Best Week in Seven Months

  Gold rallied to $4,348 — up 7.5% on the week — its strongest weekly performance since January. The collapse in rate hike expectations, combined with weakening labour data and geopolitical uncertainty, reignited the safe-haven bid. Gold had been in a downtrend since April when the peace deal first sparked a risk-on rotation. The July payrolls report may have marked its turning point.

• 📊 Q2 GDP: 1.5% — Economy Cooling Beneath the Surface

  Q2 GDP growth came in at 1.5%, down from 2.1% in Q1, confirming the economy is decelerating. However, private domestic final sales rose 3.9%, suggesting underlying demand remains resilient even as headline growth slows. The combination of slowing growth, negative payrolls, and strong corporate earnings creates an unusual environment — one where the economy is cooling but businesses are still generating record profits.

👀 Signals I'm Watching

• 🏛 The Rate Hike Is Dead — For Now
  Friday’s payrolls report effectively killed the September rate hike. Futures now price a less than 20% chance of any hike this year — down from 67% just six weeks ago. If the August CPI report (due September 10) shows inflation continuing to decline alongside oil, the conversation could shift entirely from hikes to holds — and eventually, to the first cut since the war began. New Chair Warsh’s next FOMC meeting on September 16–17 will be the test.

• 📈Tom Lee: S&P 500 to 7,900–8,000 This Month
  Fundstrat’s Tom Lee — one of the most accurate strategists of 2024–2025 — told CNBC he expects the S&P 500 to reach 7,900–8,000 by month-end. That would represent roughly 2–3% upside from current levels. His reasoning: negative payrolls remove the rate hike risk, corporate earnings are growing 25%+, and the AI trade has broadened enough to sustain the rally even if a few names pause.

• 🤖Palantir vs Meta: The AI Software Divide Is Widening
  In the same earnings season, Palantir grew revenue 93% with expanding margins while Meta saw EPS decline 14% with free cash flow collapsing. Both are “AI stocks.” Both are spending heavily. The difference is that Palantir is monetising AI through government and enterprise contracts with direct ROI, while Meta is still building infrastructure with no clear near-term return. The market is aggressively pricing this distinction — Palantir is up nearly 70% since its Q1 report while Meta has fallen roughly 25% from its spring highs.

• 🥇Gold’s Reversal Could Signal a Regime Shift
  Gold’s 7.5% weekly surge — its best since January — came alongside record equity highs, which is unusual. Normally, gold and stocks move inversely. When both rally simultaneously, it typically signals that markets are pricing in both growth (stocks) and uncertainty (gold). If gold continues to rally from here, it could indicate that the “bad news is good news” trade has a shorter shelf life than the market thinks — because eventually, bad economic news becomes just… bad.

$AXTI return over the past five days

💡In my most recent deep dive, I analyzed AXT Inc. ($AXTI) — one of my favorite under-the-radar AI stocks in the market right now. It has quickly become one of my biggest and best-performing positions.

Investment Club members were the first to know about this new addition — with full access to the deep dive, my thesis, and my price target.

If you'd like to follow along in real time, you can join us here.

George

⚠️ Red Flag to Note

The Economy Lost Jobs. The Market Hit a Record. Something Doesn’t Add Up.

The S&P 500 closing at an all-time high on the same day the economy reports its first job losses in over three years is a disconnect that deserves careful thought. The market’s logic is clear: weak jobs means no rate hike, which means lower discount rates, which means higher stock prices. But that logic has limits. If July’s job losses are the start of a trend — not a one-off distortion — then the earnings growth that has powered this entire rally is at risk. Companies can’t keep growing revenue at 25% if consumers are losing jobs and pulling back spending. The next two payrolls reports (August and September) will determine whether July was an anomaly or a warning. For now, the market is betting on anomaly. If it’s wrong, the repricing could be severe.

🔍 Insider Transactions I’m Watching

📬 Closing Note

The economy lost jobs. The stock market hit a record. Gold had its best week in seven months. And Palantir’s CEO called his quarter “otherworldly.” If that doesn’t capture the contradictions of 2026, nothing does.

The market’s logic this week was clear and internally consistent: weak jobs kill the rate hike, which lowers the cost of capital, which makes future earnings worth more today. It’s a perfectly rational trade — as long as the weakness stays contained.

But here’s the question nobody on Wall Street wants to ask: what if the job losses aren’t a one-off? What if July is the beginning of a real cooling in the labour market, not just a statistical blip? If that’s the case, the 25% earnings growth rate that has powered this entire rally is at risk — because you can’t have record corporate profits forever in an economy where people are losing their jobs.

For now, the data says this is a soft landing. GDP growth of 1.5% is slow but positive. Private demand is still growing at 3.9%. And corporate earnings — led by Palantir’s 93% revenue growth and Microsoft’s $678 billion backlog — are as strong as they’ve ever been.

But the margin for error is thinner than it’s been all year. The S&P 500 at 21x forward earnings needs everything to go right. The next payrolls report, the next CPI report, and the next Fed meeting will determine whether this is the start of a new leg higher — or the top.

Stay patient. Stay selective. And let the data guide the story.

Until next Sunday —

George ☕

Wall Street is panicking over cloud capital expenditures again.

Every time Microsoft, Amazon, Alphabet, or Meta print another multi-billion-dollar capex bill, consensus analysts scramble to downgrade stocks, questioning whether tech giants will ever recoup their massive investments. But as I'm watching the numbers roll in, I'm convinced Wall Street is looking at this through the completely wrong lens.

The current future of AI investing 2026 isn't about traditional software monetization cycles: it is a foundational industrial revolution. While software margins fluctuate quarter to quarter, the physical infrastructure layer is locking in permanent multi-year moats. Short-term analyst anxiety is creating an asymmetric buying opportunity for long-term growth investors who know where to look.

📈 The $2.4 Trillion Reality Check

When tracking the AI growth stock outlook, the sheer scale of capital spending often triggers sticker shock. Consensus estimates place cumulative AI infrastructure capex at roughly $2.4 trillion to $3 trillion through the late 2020s.

To put that in perspective, let’s look at how current projections break down across the major layers of the AI ecosystem:

Unlike the dot-com bubble, where capital was poured into speculative fiber-optic cable with zero end-users, today's multi-trillion-dollar deployment is backed by surging enterprise utilization and locked-in cloud demand. As we explore in our deep dives on The Latte, capex currently sits at roughly 0.8% of global GDP, leaving plenty of runway before reaching the historical 1.5% peaks of prior tech booms.

🔍 Why the Physical Layer Wins the Long Game

Software margins can compress, enterprise budgets can pause, and applications can face pricing pressure, but physics cannot be rewritten. The true bottlenecks in this super-cycle are not algorithms; they are megawatts, thermal dissipation, and optical bandwidth.

That is why companies operating at the physical hardware and power interface command such resilient pricing power.

• Power is the new currency: Data center power density requirements have skyrocketed, forcing hyperscalers into direct deals with gas turbine and nuclear providers: a trend we covered extensively in our breakdown of The Gas Turbine Gold Rush.

• Cooling is non-negotiable: Liquid cooling and advanced thermal management are no longer optional upgrades; they are mandatory for multi-megawatt AI clusters.

• Interconnect speed dictates ROI: As clusters scale past 100,000 GPUs, network latency between chips matters more than raw compute alone.

When you examine firms solving these exact physical bottlenecks — such as Vertiv $VRT ( ▼ 1.17% )  in thermal management or Credo $CRDO ( ▲ 0.2% )  in optical interconnects — the growth thesis becomes crystal clear.

⚠️ Navigating the Risks Without Falling for the Noise

Every major capital cycle carries legitimate risks. Ignoring them is how portfolios get blindsided during macroeconomic shifts.

Here are the key variables I'm keeping on my radar:

• Capex Overshoot & ROI Timing: If enterprise GenAI revenue realization lags behind hardware depreciation schedules, tech giants could experience temporary margin compression.

• Grid Interconnection Backlogs: Power queue delays can push revenue recognition out by quarters, testing the patience of short-term holders.

• Valuation Discipline: Not every company trading under an "AI banner" deserves a premium multiple. You must separate narrative-driven hype from cash-generative engineering moats.

This is precisely why our approach at The Latte centers on rigorous stock market research rather than headline chasing. By focusing on cash conversion, contracted order backlogs, and structural demand, we help investors navigate volatility with quiet confidence.

🚀 Positioning Your Portfolio for the Next Phase

Wall Street's obsession with next-quarter earnings misses the forest for the trees. The $2.4 trillion AI buildout is re-architecting global infrastructure for the next decade.

To position effectively right now:

• Focus on picks and shovels: Prioritize semiconductor, optical, and power infrastructure players where cash flow is secured by multi-year vendor contracts.

• Diversify beyond mega-caps: Look down the capitalization ladder into under-owned ecosystem plays, like our recent analysis on Small-Cap AI Stocks.

• Keep risk controls tight: Maintain a balanced asset allocation while letting your high-conviction growth holdings compound.

If you want to stay ahead of these macro shifts and receive our curated deep dives directly in your inbox, take a moment to subscribe to The Latte or explore our upgrade options for full access to our institutional-grade stock research.

Markets will continue to fluctuate, and the media will keep debating whether the bubble is about to burst. But as long as the infrastructure demand signals remain this robust, I'll be staying focused on the builders.

Stay disciplined,

George

Disclaimer: This post is for informational purposes only and does not constitute financial advice. Always conduct your own stock market research or consult with a financial advisor before making any investment.

For most investors, the AI infrastructure trade begins and ends with GPUs. NVIDIA $NVDA ( ▲ 0.57% ) dominates the headlines, hyperscalers dominate capital spending, and networking companies receive most of the attention whenever another trillion dollars is committed to AI infrastructure. Yet every AI cluster depends on something far less glamorous.

As AI models become larger and data centers continue scaling, moving enormous amounts of data between thousands of GPUs has quietly become one of the industry’s biggest bottlenecks. Faster processors are only useful if information can travel between them just as quickly. That is creating an unprecedented wave of investment in optical networking—a critical layer of AI infrastructure that receives only a fraction of the attention. One of the biggest beneficiaries of this trend is AXT $AXTI ( ▲ 1.12% ) .

Following another record quarter, management reiterated that demand continues to exceed production capacity, signed additional long-term supply agreements with major customers and expects revenue to continue increasing throughout the remainder of 2026 and into 2027.

Despite rallying 3,000% over the past year, the market is still underestimating how strategically important AXT’s products are becoming as AI infrastructure spending accelerates worldwide. Let’s dive in.

$AXTI ( ▲ 1.12% ) spectacular rally over the past twelve months

Portfolio Vs. The Market Performance

📈Over the past week, my 25-stock portfolio returned over 12% — compared to roughly 6% for the Nasdaq. Nearly 2x higher return, without leverage.

Every position I hold — plus the thinking behind each one — is shared with Investment Club members in real time. If you'd like to follow along, you can join us here.

George

I’m watching the AI infrastructure trade evolve from pure silicon to raw electrons. The bottleneck isn't just GPUs anymore — it's the electrical grid. As hyperscalers race to build 100,000-GPU clusters, traditional utility interconnection queues are stretching to five, six, and seven years.

Tech giants can’t wait half a decade for transmission upgrades. They are bypassing the grid entirely by securing gas turbines, natural gas microgrids, and on-site generation. Welcome to the gas turbine gold rush: a structural shift in on-site power generation AI investing that is reshaping industrial hardware markets.

📉 The 7-Year Grid Queue Bottleneck

The math is unforgiving. AI data centers require massive, 24/7 firm power. Unlike consumer internet traffic, large-scale model training runs at high load factors and cannot tolerate rolling brownouts or sudden frequency drops.

Yet, regional transmission organizations are overwhelmed. Interconnection queues feature wait times that stretch well past mid-decade.

• Grid Delays: Transmission upgrades take years of permitting, environmental reviews, and steel procurement.

• Hyperscale Urgency: Training schedules for next-generation models demand megawatts today, not in 2032.

• The Bridge Solution: On-site gas turbines can be permitted, delivered, and operational in months instead of years.

AI datacenter power generation

This mismatch creates an unprecedented corporate land grab. Tech companies are no longer passive utility customers — they are becoming off-grid power plant operators. Industry analyses suggest that more than 25% of new data centers above 500 MW will generate their own power behind the meter by 2030, up from roughly 1% today. If you want to understand where the real cash flow is heading in these new rules of AI investing, follow the turbine orders.

🔍 The Pure-Play Beneficiaries: GE Vernova $GEV ( ▲ 0.92% )  and Siemens Energy

When capital floods into physical power equipment, industrial heavyweights reap the rewards. Demand for heavy-duty and aeroderivative gas turbines has far outstripped manufacturing capacity, pushing OEM order books out to the end of the decade.

GE Vernova stands out as the most direct bellwether for AI data center gas turbine stocks. Management notes that approximately one-fifth of its gas power order book now links directly to data center and AI applications.

GE Vernova

Prices for these massive machines have climbed significantly over the past three years. When Microsoft $MSFT ( ▼ 1.64% )  or Amazon $AMZN ( ▼ 2.02% )  secures multi-gigawatt power purchase agreements in Texas or Ohio, GE Vernova’s aeroderivative units (like the LM2500XPRESS) and heavy-duty combined-cycle systems provide the firm, around-the-clock foundation required to keep training clusters humming.

⚡ Modular Power and the Off-Grid Playbook

It isn't just massive combined-cycle power plants driving the market. Speed-to-deploy dictates winners and losers in the AI race.

Elon Musk’s ventures and other nimble operators have famously bypassed traditional utility timelines by deploying modular, truck-mounted gas turbines and high-capacity natural gas engines. Caterpillar’s $CAT ( ▼ 0.06% )  Solar Turbines division and similar mobile generation suppliers have seen surging demand for small modular units that can be brought online in weeks.

This creates a brilliant "bridge-then-backup" operational model:

• Phase 1: Bridge Power: Deploy on-site gas turbines to provide primary power immediately while waiting for utility interconnection.

• Phase 2: Transition & Islanding: Once the grid connection eventually clears, shift turbines to backup emergency standby and peak shaving.

• Phase 3: Ancillary Monetization: Export surplus capacity and provide frequency regulation services back to the local grid during low-training loads.

This optionality protects initial capital expenditures and turns a temporary fix into a permanent, cash-generating grid asset.

⚠️ Risks Worth Monitoring

I remain constructive on industrial power equipment, but prudent risk management requires keeping an eye on potential headwinds:

• Levelized Cost Pressures: Natural gas microgrids and turbines can carry higher operating costs than cheap industrial base-load grid power when available.

• Environmental & Regulatory Scrutiny: Emissions standards and carbon policies could introduce friction or compliance costs for fossil-fuel-backed campuses.

• Early Component Wear: Continuous high-load cycling for AI workloads can accelerate maintenance intervals and warranty stress for OEMs.

Rising Global Gas Turbine Demand

Despite these considerations, the overarching macroeconomic reality remains clear: the data center power deficit is real, and the electrical grid cannot patch the gap fast enough.

For growth investors looking beyond software and GPUs, industrial turbine leaders offering multi-year order visibility represent a compelling, cash-generative corner of the AI super-cycle. Keep your risk controls tight, stay diversified across the supply chain, and watch how hyperscalers secure their electrons.

If you want to receive our deep-dive stock research and curated market updates directly in your inbox, subscribe to our newsletter or explore our membership upgrade options for full portfolio insights.

Until next time, stay sharp.

George

Disclaimer: This post is for informational purposes only and does not constitute financial advice. Always conduct your own stock market research or consult with a financial advisor before making any investment.

I’m watching how the hardware layers of the artificial intelligence boom continue to capture immediate, hard cash flow, long before software monetization fully catches up. While software valuations dominate headline debates, the underlying machinery is experiencing historic demand acceleration.

TrendForce’s latest data puts global AI server shipment growth on track for roughly 28% to 31% year-over-year expansion. Behind this massive surge lies a 40% plus jump in capital expenditures from top Cloud Service Providers (CSPs) like Microsoft $MSFT ( ▼ 1.64% ) , Amazon $AMZN ( ▼ 2.02% ) , and Meta $META ( ▲ 0.7% ) .

If you are hunting for AI server stocks to watch 2026, the hardware supply chain is where cash conversion meets undeniable secular momentum. Let's break down why this infrastructure wave is translating into exceptional growth opportunities and three specific hardware leaders positioned to capture the upside.

📈 The Macro Catalyst: Why CSP Capex is Accelerating

When major tech giants report earnings, Wall Street often fixates on quarterly profit margins. I prefer to look further down the income statement at capital expenditure guidance.

Recent earnings reports confirm that hyperscalers aren't pulling back on infrastructure: they are doubling down.

Hyperscaler Capex growth

• Unprecedented Scale: Top North American CSPs are projected to boost combined capex significantly in 2026, funding both new rack-scale AI clusters and the retirement of legacy 2019-era cloud hardware.

• The Utilization Shift: AI inference workloads are scaling faster than training workloads, requiring dense, liquid-cooled server racks that only specialized OEMs and networking giants can deliver.

• Supply Chain Bottlenecks: High-bandwidth memory, advanced packaging, and optical networking components remain tight, giving pricing power back to tier-one hardware suppliers.

As I noted in our recent deep dive on Lam Research $LRCX ( ▼ 0.96% ) , semiconductor equipment and hardware manufacturing form the absolute bedrock of the modern tech economy. When demand outpaces supply at this magnitude, tier-one server assemblers reap immediate rewards.

🔍 The Server Supply Chain Playbook

Investing in AI hardware requires understanding where the economic value concentrates. It isn't just about raw chip design; it's about system integration, thermal management, and ultra-low latency networking.

AI supply chain leaders

When evaluating AI server stocks to watch 2026, I look for three distinct operational advantages:

1. Rack-Scale Customization: Ability to design and ship complete liquid-cooled rack solutions directly to data center operators.

2. Enterprise Distribution: Deep relationships with Fortune 500 companies transitioning their on-premise infrastructure into hybrid AI clouds.

3. High-Speed Networking Backbone: Proprietary switching technology that prevents massive GPU clusters from choking on data transit.

For investors seeking structured, regular updates on these infrastructure trends, our curated archive tracks every major quarterly shift in the growth stock landscape.

🚀 3 Hardware Stocks Riding the Trend

1. Super Micro Computer $SMCI ( ▼ 2.99% )  

Super Micro remains the undisputed pioneer in modular, liquid-cooled server architecture. By partnering closely with NVIDIA $NVDA ( ▲ 0.57% ) , AMD $AMD ( ▲ 2.19% ) , and Intel $INTC ( ▼ 0.05% ) , SMCI has mastered the art of delivering custom rack-scale solutions faster than legacy competitors. While margins face periodic scrutiny, their top-line revenue expansion proves that hyperscalers rely heavily on their time-to-market agility.

2. Dell Technologies $DELL ( ▲ 1.73% )  

Dell is quietly executing one of the most successful enterprise pivots in tech history. Their PowerEdge server lineup, optimized for high-density AI inference workloads, has secured massive commitments from both sovereign cloud projects and tier-two neocloud providers. Dell's unmatched global supply chain and enterprise service network give it a distinct moat against pure-play startups.

3. Arista Networks $ANET ( ▲ 2.68% )  

You cannot talk about AI servers without addressing the networking fabric holding them together. As GPU clusters scale into tens of thousands of chips, standard Ethernet switching hits severe bottlenecks. Arista’s ultra-low latency routing and optical switching platforms make them the indispensable tollbooth for modern AI data centers. Their gross margins and consistent free cash flow generation set a gold standard in hardware investing.

⚠️ Managing Risk in a Fast-Moving Hardware Cycle

Every growth cycle carries cyclical risks worth monitoring. Hardware manufacturing is inherently capital-intensive, and component pricing can fluctuate rapidly based on supply chain adjustments.

Hardware stock portfolio

• Keep risk controls tight: Position sizes should reflect the cyclical volatility typical of technology hardware equities.

• Watch inventory turns: Monitor quarterly inventory growth relative to revenue to ensure customer demand remains robust.

• Stay diversified: Pair high-growth hardware plays with steady infrastructure providers, similar to the grid-backed opportunities we explored in our Forgent Power Solutions $FPS ( ▲ 9.5% )  analysis.

Looking Ahead

The 31% surge in AI server shipments isn't an isolated statistical blip: it is the tangible manifestation of a multi-year infrastructure super cycle. By focusing on execution-driven hardware leaders like SMCI, Dell, and Arista, investors can capture the most lucrative phase of the AI buildout.

If you want to stay ahead of these market movements without drowning in noise, consider joining our community. You can subscribe here to receive regular, simplified deep dives directly in your inbox.

Until next time, keep your risk controls tight and your perspective long-term.

George

Disclaimer: This post is for informational purposes only and does not constitute financial advice. Always conduct your own stock market research or consult with a financial advisor before making any investment.

With hyperscaler capital expenditures projected to breach $750 billion over the next few cycles, the narrative on Wall Street is shifting. While retail portfolios remain heavily concentrated in mega-cap bellwethers like Nvidia $NVDA ( ▲ 0.57% ) , Microsoft $MSFT ( ▼ 1.64% ) , and Alphabet $GOOG ( ▼ 1.24% ) , the real asymmetric upside in this multi-year boom is migrating downstream.

If you are scanning for small cap AI stocks to watch, you are looking beyond the front-end software layer into the gritty, essential machinery of the hidden AI infrastructure stocks: the power providers, thermal architects, and specialized engineers making the data center revolution possible.

Let’s look at what is driving this rotation, and four small-cap players positioned to capitalize on the bottlenecks.

⚡ The Infrastructure Bottleneck: Power, Heat, and Lead Times

The constraint on artificial intelligence is no longer algorithmic imagination: it is sheer, unadulterated physics.

Data center construction lead times have stretched to an agonizing 53 months in major tier-one markets. Power grids are groaning under the weight of cluster demands that rival small cities, and the cooling bottleneck has turned thermal management from an engineering afterthought into a board-level imperative.

When hyperscalers need gigawatts yesterday, they cannot wait for traditional utility pipelines. They need modular, high-density, rapidly deployable solutions. That reality creates fertile ground for nimble, under-the-radar innovators.

1. Applied Digital $APLD ( ▼ 4.72% ) : The Data Center Powerhouse at a Discount

Applied Digital has quickly transformed from a specialized high-performance compute host into a premier designer and operator of next-generation data centers tailored specifically for AI workloads.

• The Catalyst: A massive $36 billion contract backlog anchored by top-tier hyperscale customers validates their execution capability.

• The Numbers: Revenue has surged roughly 400% year-over-year, illustrating explosive commercial traction.

• The Setup: Despite surging operational milestones, the stock currently trades at roughly half of its 52-week high, presenting a compelling entry window for growth investors willing to look past near-term capital expenditure jitters.

Applied Digital represents the classic "picks and shovels" play for institutional-grade AI hosting.

2. Nano Nuclear Energy $NNE ( ▼ 0.69% ) : Powering the Next-Gen Grid

As AI data centers demand dedicated, zero-carbon power sources directly on-site, nuclear energy is experiencing an unprecedented renaissance. Nano Nuclear Energy is pioneering the development of portable, pre-revenue nuclear microreactors.

• The Catalyst: Approaching regulatory milestones and strategic pilot partnerships designed to place clean, uninterrupted baseload power directly adjacent to high-density compute facilities.

• The Profile: As a pre-revenue development-stage company trading around the $30 mark, NNE is undeniably speculative.

• The Thesis: For investors seeking asymmetric moonshot exposure to the intersection of clean energy and AI, Nano Nuclear offers pure-play optionality on microreactor commercialization.

3. TSS Inc. $TSSI ( ▼ 3.43% ) : Solving the Thermal Crisis

You can build the most advanced GPU cluster in the world, but without sophisticated rack integration and direct liquid cooling, it will thermal-throttle within seconds. Enter TSS Inc., a small-cap specialist in data center integration and thermal management.

• The Catalyst: Direct liquid cooling (DLC) has transitioned from a nice-to-have feature to an absolute requirement for Blackwell and next-gen accelerator architectures.

• The Setup: TSSI operates as the quiet connective tissue between semiconductor manufacturers and hyperscale facility operators, ensuring custom rack configurations are deployed flawlessly and on schedule.

• The Outlook: As cooling bottlenecks intensify across the industry, micro-cap and small-cap suppliers like TSS stand to capture outsized operating leverage.

4. AirJoule Technologies $AIRJ ( ▼ 2.17% ) : Sorption-Based Cooling Innovation

Cooling data centers traditionally consumes staggering amounts of water and electricity. AirJoule Technologies is targeting this exact vulnerability with its proprietary sorption-based cooling and water-harvesting systems.

• The Catalyst: Trading under $5, this microcap innovator is scaling atmospheric water harvesting and advanced thermal management specifically engineered to reduce the water usage effectiveness (WUE) of modern AI facilities.

• The Thesis: While early-stage and highly volatile, AirJoule addresses one of the most critical environmental and operational hurdles facing data center developers in arid regions. Keep risk controls tight, but keep this one on your watchlist.

📊 Small-Cap AI Infrastructure Comparison Matrix

🔍 Navigating the Small-Cap Landscape: Final Thoughts

Investing in small-cap growth stocks requires a stomach for volatility. Macroeconomic crosscurrents, customer concentration risks, and capital raise requirements can test even the most disciplined investor.

However, when an entire technological paradigm shifts: as we are witnessing with the $750B hyperscaler capex supercycle: the greatest alpha is rarely found where everyone else is looking. By monitoring these hidden infrastructure plays, staying diversified, and sizing positions prudently, you can position your portfolio ahead of the curve.

Stay disciplined, keep your risk controls tight, and I’ll see you in the next issue.

George

Disclaimer: This post is for informational purposes only and does not constitute financial advice. Always conduct your own stock market research or consult with a financial advisor before making any investment.

👋 ICYMI

Four Magnificent Seven companies reported earnings this week. The results were historic — in both directions.

Microsoft $MSFT ( ▼ 1.64% ) surged 16% on Thursday — its biggest single-day rally ever — adding roughly $450 billion in market capitalization in a single session, the largest one-day value creation in stock market history. The catalyst: a $678 billion backlog, Azure growth that beat every estimate, and a clear demonstration that AI spending is translating into enterprise revenue at scale.

Amazon $AMZN ( ▼ 2.02% ) rallied 10% after hours on Thursday after reporting AWS growth of 37% — its fastest cloud expansion since 2021 — alongside commentary that AI demand is outstripping capacity.

Meta $META ( ▲ 0.7% ) fell nearly 10% after reporting a 14% decline in earnings per share and free cash flow collapsing to just $784 million — vindicating the market's fear that not all AI spenders are AI earners.

Apple $AAPL ( ▼ 0.52% ) fell over 7% on Friday despite beating on both revenue and EPS — the selloff was driven by weak Q4 guidance as CEO Tim Cook flagged an "increasing impact" from a global memory shortage that is constraining iPhone, iPad, and Mac supply.

The S&P 500's IT sector rallied 5% on Thursday alone — the largest single-day gain for the tech sector since mid-2025. The S&P 500 closed at 7,489 and the Nasdaq at 25,373. South Korea's KOSPI surged 17.91% on Friday — the largest single-day gain in its history, as Microsoft's results sent semiconductor and AI-linked stocks soaring across Asia.

The Fed held rates steady at 3.50–3.75% on Wednesday, as expected. But the 30-year Treasury yield spiked above 5.2% — its highest since 2007 — as investors worried the central bank is falling behind on inflation. Fed funds futures now price in a quarter-point hike in September.

🔁 Market Movers

• 💻 Microsoft +16%: The Biggest Day in Stock Market History

  Microsoft's post-earnings rally added approximately $450 billion in market capitalization — the largest single-day value creation any public company has ever produced. Azure growth beat every estimate. The company disclosed a $678 billion commercial backlog. And CEO Satya Nadella demonstrated, with hard numbers, that Microsoft's $190 billion AI capex plan is generating real, scalable enterprise revenue. The market had spent weeks punishing AI spenders. Microsoft answered with the most convincing proof-of-concept in AI history.

• 🛒 Amazon +10%: AWS Growth Fastest Since 2021

  Amazon reported after Thursday's close with AWS cloud revenue growing 37% — a dramatic acceleration from recent quarters and the fastest growth rate since 2021. Management said AI demand is exceeding capacity, necessitating continued infrastructure buildout. The stock jumped 10% after hours. Combined with Microsoft's results, the message is clear: cloud-native AI platforms with direct enterprise monetization are the winners of this cycle.

• 📱 Meta -10%: Free Cash Flow Collapse Validates the Bears

  Meta reported a 14% decline in EPS and free cash flow of just $784 million — a dramatic deterioration from its typical multi-billion-dollar quarterly cash generation. The Meta Compute cloud business, which drove a 15% rally two weeks ago, has not yet materialized into revenue at a meaningful scale. The stock dropped nearly 10%. For Meta, the AI spending story has become a cash flow problem until proven otherwise.

• 🍎 Apple -7%: Record Quarter, But Supply Constraints Spook the Market

  Apple reported its best June quarter ever — $109.4 billion in revenue (+16% YoY), EPS of $2.02 (+29% YoY), iPhone revenue up 22%, and Mac up 29% (CNBC). But the stock fell over 7% on Friday after Q4 guidance of 9–11% revenue growth missed the 12% consensus. CEO Tim Cook warned that a global memory shortage — driven by surging AI demand for DRAM and NAND — is creating supply constraints that will "increase significantly" in the September quarter. Services revenue of $30.7 billion missed estimates, and Greater China came in below expectations at $18.8 billion.

• 🏦 Fed Holds — But Yields Scream Higher

  The Fed held rates as expected, but the 30-year Treasury yield spiked above 5.2% — its highest closing level since 2007. Markets are now pricing a quarter-point hike in September. Chair Warsh gave no indication of urgency to cut despite softening economic data. The yield curve's message is clear: the bond market believes the Fed is behind the curve on inflation.

• 🇰🇷 KOSPI +17.91%: Largest Single-Day Gain in History

  South Korea's stock market surged nearly 18% on Friday — the biggest single-day percentage gain in the KOSPI's history — driven by Microsoft's AI validation and its implications for semiconductor demand from Samsung and SK Hynix. The move underscores how global the AI investment thesis has become — one American company's earnings report moved an entire nation's stock market by 18%.

👀 Signals I'm Watching

• 📊 The AI Capex Debate Is Settled — For Now

  For months, the market has asked whether $600 billion+ in hyperscaler AI spending would generate returns. This week, Microsoft answered with a $678 billion backlog and the biggest single-day rally in market history. Amazon answered with 37% AWS growth. The AI infrastructure buildout isn't just hype — it's generating enterprise revenue at an accelerating rate. The companies that can prove monetization (Microsoft, Amazon) are being rewarded massively. The companies that can't (Meta, Tesla) are being punished. The debate isn't over, but the burden of proof has shifted.

• 🍎 Apple's Memory Shortage Is the AI Trade's Unintended Consequence

  The irony is striking: the same AI-driven demand for memory chips that sent Micron to $1 trillion and powered the entire semiconductor rally is now constraining Apple's ability to build iPhones, iPads, and Macs. Cook's warning that supply shortages will "increase significantly" next quarter means the AI hardware boom is cannibalizing consumer electronics supply chains. This is a risk the market hadn't fully priced — and it extends beyond Apple to any hardware company that competes for DRAM and NAND with hyperscale AI data centers.

• 📈 The September Rate Hike Is Now Consensus

  Fed funds futures are pricing a 25 basis point hike in September — a dramatic shift from just months ago when the debate was about cuts. The 30-year above 5.2% is already tightening financial conditions without a formal hike. If the July CPI report (due mid-August) shows inflation accelerating again, the September hike becomes a near certainty. If inflation cools as oil falls, Warsh may have room to delay.

Most recent deep dive and stock performance

⚡️ In my most recent deep dive, published on July 29, I covered Bloom Energy's ($BE) earnings results and explained why they reaffirmed the company's bull case — making the stock look unjustifiably cheap. On July 30, $BE surged as much as 30% as investors reacted to its blockbuster earnings and guidance.

Investment Club members got my full thesis, portfolio update, and an updated price target on the stock. If you'd like to follow along in real time, you can join us here.

George

⚠️ Red Flag to Note

The Market Is Increasingly Dependent on Two Companies

Microsoft and Amazon now account for an outsized share of the S&P 500's year-to-date performance — and this week dramatically widened the gap. Microsoft's $450 billion single-day addition is larger than the entire market capitalisation of 95% of S&P 500 companies. When one company's earnings report can move the KOSPI 18% and add more value in six hours than most companies create in a lifetime, the market's concentration risk has reached levels that warrant genuine caution. The AI thesis is real. The earnings are real. But the degree to which global equity markets now depend on the execution of two or three companies is historically unprecedented.

🔍 Insider Transactions I’m Watching

📬 Closing Note

This was the week that proved AI is real. Not as a narrative. Not as a promise. As revenue. As backlog. As the single largest day of value creation in stock market history.

Microsoft's $678 billion backlog answered every sceptic who questioned whether $190 billion in AI spending would generate returns. Amazon's 37% AWS growth showed that cloud demand isn't just holding — it's accelerating. And the KOSPI's 18% single-day surge demonstrated that the AI thesis is now a global investment story, not just an American one.

But the week also showed what happens when you can't prove it. Meta lost 10% because free cash flow collapsed. Tesla $TSLA ( ▼ 0.67% ) is down 19% in two weeks because spending is rising faster than revenue.

The market has drawn its line in the sand: show me the revenue, or I'll show you the door.

But here's what I keep coming back to: corporate America is earning more than it ever has. The AI buildout is real and accelerating. And the companies that can prove monetization are being rewarded at historic levels.

The second half won't be easy. But for investors who can distinguish between AI earners and AI spenders — between backlog and hope — it could be the most rewarding stretch of the cycle.

Stay patient. Stay selective. And let the data guide the story.

Until next Sunday —

George ☕

If you spent the first half of 2026 glued to Nvidia’s $NVDA ( ▲ 0.57% )  ticker, you might have walked away feeling a bit underwhelmed. In a year when the undisputed poster child of the generative AI boom barely budged — logging modest low-teens gains — the real market fireworks were happening elsewhere.

Wall Street didn't abandon the AI trade. Instead, investors broadened their scope, sparking an aggressive rotation into physical infrastructure, memory, and foundry turnarounds. If you were searching for the best AI stocks to buy now in 2026, the winners weren't always the usual mega-cap suspects.

Let's look at the scoreboard and unpack why memory and CPUs stole the show in H1.

📉 The H1 2026 AI Stock Scorecard

The narrative that "Nvidia is the entire AI trade" officially fractured over the last six months. While Nvidia consolidated its multi-trillion-dollar gains, suppliers addressing critical physical bottlenecks surged hundreds of percent.

🔍 Memory Scarcity Takes Center Stage

Memory chip stocks stole the show

I’m watching memory markets closely right now, and for good reason. As AI models scale past parameters that fit neatly into standard GPU memory pools, high-bandwidth memory (HBM) has become the ultimate chokepoint in the datacenter.

Micron $MU ( ▲ 0.39% )  didn’t just participate in this cycle; it dominated it. With HBM supply completely booked out through late 2027 and well into 2028, pricing power shifted decisively toward manufacturers.

• Pricing power is sticky. Unlike commoditized DRAM cycles of the past, AI-grade memory commands structural margin expansion.

• The beta trade. Micron acted as a high-beta play on infrastructure scarcity, rewarding patient investors who looked past software hype.

As I noted in our recent deep dive on Lam Research $LRCX ( ▼ 0.96% ) , wafer fabrication equipment and advanced packaging are where the rubber meets the road. Memory isn't just an auxiliary component anymore: it's the core limiter of inference speed.

⚡ The CPU Resurgence and Foundry Turnaround

For the past three years, the market treated traditional server CPUs as legacy tech destined to stagnate. Then came 2026.

Intel surprised skeptics by proving that data center AI isn't only about GPUs: it's about orchestrating massive heterogeneous workloads. Intel’s Data Center and AI segment jumped over 20% year-over-year, providing the foundational compute backbone that complementary accelerators plug into.

• The re-rating catalyst. When a legacy heavyweight trades at depressed valuations and delivers triple-digit gains, it signals a massive sentiment shift.

• Supply chain rebalancing. Buyers are actively diversifying supply chains away from single-vendor bottlenecks, creating breathing room for turnaround plays.

It’s a classic reminder: in cyclical growth investing, you never want to write off an asset when consensus sentiment reaches peak pessimism.

⚠️ The Great Rotation: Hardware vs. Software

Surging CPU demand

While hardware and memory suppliers enjoyed a banner first half, high-multiple software names faced a reality check. Palantir $PLTR ( ▼ 0.43% )  and other pure-play AI software darlings experienced meaningful pullbacks — some dropping over 30% — as Wall Street demanded proof of monetization matching the lofty valuations.

This wasn't a rejection of enterprise AI adoption. It was a healthy digestion phase.

• Infrastructure comes first. You cannot deploy autonomous agentic workflows or heavy enterprise analytics without reliable power, cooling, and high-speed memory.

• Valuation discipline matters. Markets are currently rewarding cash-flowing physical assets over speculative future software multiples.

This echoes what we saw across the power grid landscape. As highlighted in our analysis of Bloom Energy $BE ( ▲ 0.9% ) , physical constraints dictate the speed of digital progress.

🔭 Looking Ahead to H2 2026

Investment roadmap

So, how should we position our portfolios as we head into the second half of the year?

First, keep risk controls tight. Multi-hundred-percent runs in names like Micron and Intel naturally invite near-term volatility and profit-taking. Second, don't mistake a quiet stock price for a broken business: Nvidia’s underlying cash flows and GPU dominance remain entirely intact, even if its stock took a breather to digest prior gains.

The overarching takeaway from H1 2026 is simple: the AI trade has matured. The era of blindly buying any stock with "AI" in the press release is over. Today, sustainable outperformance requires identifying where the next physical bottleneck lies: whether in memory, advanced packaging, or power generation.

Want to stay ahead of these sector rotations before they hit mainstream headlines? Make sure you're subscribed to The Latte to get our institutional-grade research delivered straight to your inbox. If you're ready to unlock our full high-conviction growth portfolio and early-access deep dives, consider upgrading to The Latte Premium.

Stay diversified, keep your research rigorous, and I'll see you in the next issue.

George

Disclaimer: This post is for informational purposes only and does not constitute financial advice. Always conduct your own stock market research or consult with a financial advisor before making any investment.