This is a longer letter. A few events in succession are worth covering. Privacy has been under the gun the past few weeks and the general trend –– personal privacy is degrading and risks are increasing.

Age Verification Wars

The UK law requiring age verification went into effect this week. Websites with adult content (pornography, but also sites with NSFW content like Reddit) will have to comply by verifying users are not minors. Immediately, VPN signups soared as people looked to spoof their IPs to avoid verification.

There are arguments that this is necessary for shielding an extremely online youth population from age-inappropriate content, but in practice, age verification means uploading photos, IDs, or other personal information to verify your age. Verification means users must ask themselves, “Do I trust the company to be responsible with my sensitive data?”

That trust may be easier to give to established tech giants like YouTube (which is working on their own age verification). But last week, a concerning privacy incident came out of Tea, an app advertised as "dating safety tools that protect women" where women could share dating experiences and advice, including posting about men to avoid and conduct background checks. Last week it was reported that a "hack" had revealed 70,000+ users from the Tea database. I say "hack" because the database was apparently public-facing, so really it was just found. What's worse is the leak included thousands of photos of women's faces and driver's licenses and the information was shared extensively on 4chan. A map was circulating with the addresses of those in the leak.

Tea says that they will offer personal identity help to victims. But it is hard to imagine how emotionally damaging this leak could be for those in it, as well as introducing anxiety about identity theft, stalking, or other real-world harms.

Vibe Coding Security Crisis

This leak also exposed a larger worrying trend for startups. Vibe coding is going to lead to what some are calling vibe insecurity. As people create usable products with text prompts, it is more likely that AI will make security lapses, whether that is making databases public or exposing API keys. It introduces a set of people building software that won't be familiar with their own privacy and security protocols.

AI Chatbots and Legal Exposure

In a different privacy concern, people are realizing that AI chatbots do not have any legal privileged protections. In the US your electronic records are generally considered under ECPA (Electronic Communications Privacy Act), originally passed in the '80s. People don't like to think about this, but in almost all cases, under a legal investigation if a judge signs a search warrant for information held by a tech company, the company will disclose your information. The rationale is that your electronic records are somewhat akin to notes you'd keep in your home which would also be accessible via search warrant. But I don't think a lot of people think like this. Sam Altman made this point on a recent podcast, which is especially concerning as people are using AI chatbots in sensitive ways like legal, therapy, and medicine, without the legal protection we have afforded those professions.

People might argue that this is a case of "if I am not doing anything wrong my information is not at risk," but another huge risk right now is the technological literacy of law enforcement and judges in the United States. While working in tech legal compliance, I once saw a search warrant for location data for the entire city of Virginia Beach AND a judge had approved it. That was a legally compelling warrant for the data of millions of people. In this case, counsel got involved and pushed back on the warrant, but it shows that your data could be exposed by mere proximity to an investigation or a misunderstanding from law enforcement or a judge on what information is needed.

Instagram post

The Astronomer Moment

Finally, it would be tough not to mention the most viral privacy incident recently: the Astronomer scandal. The internet exploded when the CEO and HR head were caught embracing at a Coldplay concert. The consequences were swift when the internet, late night talk hosts, sports stadiums, and newsletter writers piled on with memes and condemnation. The CEO resigned, the HR head was put on leave, and Astronomer hired Gwyneth Paltrow (Chris Martin's ex) to do a PR PSA to capitalize on the attention. It is a reminder that cameras really are everywhere and the viral machine moves quickly. With many startups hoping to put always on AI recorders on people and smart glasses with cameras making a strong comeback, we are moving towards an even more always observed world.

Startups

How will this fold into the zeitgeist of startups. On one hand, privacy concerns might open an opportunity. Who is building the Plaid of personal verification or security review for vibe-coded apps? There's a real need for infrastructure that can quickly audit AI-generated code for privacy vulnerabilities before deployment.

On the other, startups are certainly exacerbating this problem between vibe coding, AI assistants like Friend or Limitless, and always-recording smart glasses. There doesn’t seem to be a privacy backlash as people seem happy to accept the tradeoff for the advancements in technology. Which might mean startups can push the envelope even more.

Our Sponsor

Founders, get ahead of tax season: Gelt is your year-round tax partner built for startups, investors & entrepreneurs. Reduce your burden, boost returns, and make smarter tax moves—fast.

Stop overpaying or scrambling come tax season. Gelt delivers proactive strategy and full-service filing, powered by elite CPAs and a powerful platform. Trusted by founders and loved by investors, we've helped 200+ businesses save over $5M (and counting).

Check Eligibility

Startup(s) of The Week

This week we have a category of startups: Synthetic Populations

Imagine your entire life, upbringing, the world around you, was a simulation. The entire universe was generated around you, molding you into the unique person with unique preferences, all just to see whether you respond positively to Disney’s newest ad campaign for Thor 13 - Revenge of Mjölnir. Oh, and you are simulated too.

That’s the future of customer insights, campaign research, and user interviews that a growing number of AI startups want to create. The idea is that AI can generate huge and more realistic synthetic populations (based on a few real data points, user interviews, etc), which could help predict real behavior whether in response to ad campaigns, product launches, or even legal arguments in a courtroom.

Two startups to highlight in this area:

https://getpersona.xyz/ - Pre-seed stealth company.

Viewpoints.ai - Simulate large-scale interactions with text, image, and video content for real-time, accurate insights. - This startup is run by Stanford PhD, Leo Yeykelis, and is coming out of the PearX accelerator (backed by Pear VC)

Signals

Industry leaders came out with a swarm of announcements this past week. OpenAI’s re-release of Codex, Windsurf released SWE-1 models, Airbnb added new services, Notion added AI features including AI meeting notes, and Eleven Labs put out a universal soundboard.

In some ways, it shows how responsive these tech companies can be to new model developments, trends, and competition.

Open AI and Windsurf | Codex and SWE-1

OpenAI released Codex, a cloud-based software engineering agent.

Powered by a new model codex-1. OpenAI continues to push into more agentic task handling with Operator and this, where Codex can “request further revisions, open a GitHub pull request, or directly integrate the changes into your local environment”. The improvements on the current SWE-Benchmarks aren’t paradigm shifting, but OpenAI has put down the base for your eventual true AI engineer coworker.

Meanwhile, Windsurf, recently acquired for ~$3B, launched SWE-1, its software engineering-specific model. Between Windsurf’s IDE, SWE-1, and OpenAI’s Codex, the AI developer tools race (with players like Cursor, Bolt, and Replit) just got tighter.

Notion | AI Meeting Notes

Notion Introduced AI meeting notes. It is another expansion from Notion the wiki page company to Notion, the all in one business platform. It also is a clear response to Granola AI, which just raised a $43 million series B.

Eleven Labs | Universal Soundboard

Eleven Labs launched a bit of fun this week with their Universal Soundboard, expanding their audio offerings with more accessible, creative tools.

Twitter tweet

Sponsored Webinar! How ​​AI Is Reshaping Productivity - Join the Conversation on How Leaders Are Staying Ahead of the AI Shift

AI is reshaping how teams build, scale, and win—and Hubstaff is on the front lines. As a leading time tracking and productivity platform powering 112K+ remote and hybrid teams, we’re sharing what we’re seeing so you don't get left behind.

Join us live on Wednesday, May 28th (9 AM PT | 12 PM ET) for a tactical panel: “What’s Working and What’s Next in AI”. You'll hear how founders and early adopters are integrating AI into workflows, which tools are actually delivering ROI, and how leaders are identifying the winners in this fast-moving space.

Whether you're building, scaling, or backing the next big thing, this conversation offers tactical insights and future-looking perspectives you won't find elsewhere. Join us and stay ahead of the curve in the AI productivity revolution.

Save My Seat

Hi Pitch Club. This weeks startup feature is VibeCode. The company has been gaining traction on X, but is pre-public launch, and has been showcasing their Vibecode App (in beta for iOS) that lets you build other mobile apps, and can help deploy them directly to test flight.

Why we are interested

There is something tantamount to a gold rush going on in vibe-coding/IDE platforms right now. Cursor, Replit, Bolt, Lovable, Vercel, to name a few, are all competing to be the low tech knowledge platform of our dreams, turning words into products effortlessly for ~$20 a month.

So why try to back a new player? Because we don’t think anyone has really figured out mobile yet. Almost all the platforms are using similar base models, building in react native, mostly designed for web apps and websites. VibeCode (also react native I believe), is at least attempting to differentiate by directly going after the mobile application market. It is a market filled with many, many founders that are desperate willing to pay handsomely to get their MVP to the app store.

A huge part of the utility of products like Replit is that they have built deployment directly into their platforms. Making it as easy to have a live web app as clicking the deploy button – as Replit handles the build, version control, and hosting. If VibeCode as a company can accomplish a similar ease for deploying iOS apps (which requires reviews and quality controls), they will immediately attract thousands of users. There is even more opportunity for custom models for mobile development (think a Swift/Swift UI specific Cursor).

The Product

Twitter tweet

Founder: Ansh Nanda - Duke Alum, Former BlueSky, Former YC Series A

Round: Pre-seed or no current funding data available

A little note from an advertiser (in an area that I think AI has been supremely helpful) for AI meeting notes.

Attention AI enthusiasts: This podcast is for you.

In This New Way, a new podcast by Fellow, Aydin Mirzaee invites leaders to share their screens and give demos of how they’re building internal tools and reinventing work with AI.

No theory, no fluff – just tactical, actionable insights from those leading the charge.

Start listening

We hope you all have a wonderful week!

Rhian spent this weekend drafting a Request for Startups for her firm. I always love reading requests for startups because it feels like an immediate snapshot of the venture market, the zeitgeist.

What are investors desperate to get into right right now, and what do they see as the immediate future. The lists serve as soft influence for potential founders to build in investable areas and as a way to open potential funnels, especially for niche firms.

Not surprisingly, the lists are heavily focused on different AI applications or the infrastructure around it.

Requests For Startups to look through:

My Favorite Request for Startups recently is from Ben Lang at NextPlay, which you can read here, which asked a series of investors the same question as above: what problems need solving, what is the future going to look like, what should people be building?

Two examples:

1. Alex Brussell: A dating app where you get only one swipe called One In The Chamber.

2. Nitesh Banta: There should be an AI tutor for early childhood education. AI's impact on education will be profound. I'm excited about new modalities to personalize education at the earliest stages when many parents are unsure how to introduce technology to infants and toddlers.

YC’s Spring Request for Startups

• A Secure AI App Store

• Datacenters

• Compliance and Audit

• DocuSign 2.0

• Browser & Computer Automation

• AI Personal Staff for Everyone

• Devtools for AI Agents

• The Future of Software Engineering

• AI Commercial Open Source Software (AICOSS)

• AI Coding Agents for Hardware-Optimized Code

• B2A: Software Where the Customers Will All Be Agents

• Vertical AI Agents

• Startup Founders with Systems Programming Expertise

• Inference AI Infrastructure in the World of Test-Time Compute

A16z SpeedRun and A16z Big Ideas for Tech

• American Dynamism

• Bio + Health

• Consumer

• Crypto

• Enterprise + Fintech

• Games

• Growth-Stage Tech

• Infrastructure

• (Recent addition) Vibe Creation Platforms

It is Tax Seasonnnnn

This week, we have a sponsor to help.

Business Tax Filings by Firstbase - 4 Days Left to Avoid IRS Penalties

If you run a US corporation, your tax deadline is April 15.

Not ready? Firstbase is here to help and will file a 6-month extension for your company.

Avoid late penalties and buy time through October 15.

With Firstbase, you also get

• A dedicated tax expert

• Unlimited support

• Help maximizing tax credits

Get Started Today

Happy Thursday. We’ve got some interesting venture news this week—and a mostly unnecessary rant.

Around the Horn

Global Moves: Qatar's Venture Capital Ambitions

Qatar is leveraging its sovereign wealth fund to attract venture capital firms to Doha, aiming to establish itself as a regional hub. The Qatar Investment Authority (QIA) has allocated nearly half of a $1 billion fund of funds to draw more venture capital, supporting its burgeoning startup ecosystem. The move isn’t surprising. Like hosting the World Cup, this initiative is part of Qatar's broader strategy to diversify its economy and show up on the global stage.

Defense Tech Alert: Rune Emerges from Stealth

Rune, a defense technology startup founded by former Anduril employees David Tuttle and Peter Goldsborough, emerged from stealth with $6.2 million in seed funding. That’s a pretty large seed, but the round was led by Andreessen Horowitz, with participation from Point72 Ventures and XYZ Ventures, so they are already backed by big names. Rune is developing TyrOS, software designed to aid the military in managing field logistics more efficiently by autonomously tracking supplies and predicting future needs.

PE has Lowered Expectations

Private equity had high hopes for a booming 2025, fueled by anticipated regulatory relaxations and lower interest rates. However, uncertainty around interest rate cuts has led to a 53% slowdown in U.S. activity compared to the previous year. Inflation and a robust labor market suggest that rates might remain steady or even rise. Conversely, a weaker-than-expected labor market could prompt rate cuts, potentially benefiting private equity. Treasury Secretary Scott Bessent notes that the private sector is in a recession, though his job creation statistics under the current administration have been called into question. Expect similar slow activity across private markets…like venture.

Anthropic Nears Completion of $3.5 Billion Funding Round

AI startup Anthropic is finalizing a $3.5 billion funding round, potentially valuing the company at $61.5 billion. The round may include investments from prominent venture firms such as Lightspeed Venture Partners, General Catalyst, and Bessemer Venture Partners. This significant investment underscores the growing venture capital interest in AI startups. Co-founded by former OpenAI executives, including siblings Dario and Daniela Amodei, Anthropic recently launched an advanced AI model, Claude 3.7 Sonnet, aimed at enhancing response speed and reasoning capabilities.

Job Board

• Investment Principal, Global VC Fund (San Francisco, In-Office): Link

• Deep Tech Investment Intern 2025 Summer (MBAs and PhDs): Link

A message from a sponsor:

Is cold outreach feeling colder lately?

Let Redrob's AI Open Doors for You!

Redrob helps startups connect effortlessly

with top prospects, talent, and investors.

Use natural language search across 200M+ profiles—and put outreach on autopilot to collect high-value meetings.

AI so easy your parents could use it 😌

Try Redrob free for up to 6 months!

Get Started Free

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

Happy Thursday everyone.

One of the first movers in wearable AI, the Humane Pin, is officially dead. Humane agreed to sell most of its assets to HP for $116 million. There is no indication that this is the “pop” of the AI bubble, but it does show the risks for investors taking sky-high valuations (Humane raised over $230 million, ouch) and for early adopters who now own a $699, useless metal clip with a blurry camera.

Humane might have failed because it moved too early and succumbed to the pressure of making returns on their $1 billion+ valuation. But many would say that the Humane Pin was just a terrible product. Complaints cited a bad camera, huge lag in responses, hallucinations, and poor battery life, with Youtuber MKBHD saying it was the “Worst Product I’ve Ever Reviewed”. 

Maybe some of the other physical AI assistants will get it right, players like limitless.ai or friend.com have similar offerings. I genuinely hope they do. The idea of an always on AI a-la the movie Her, strikes some as dystopian, but executed correctly it might be less dystopian than our current reliance on screens and smartphones, constantly designed to hook our attention with slot-machine-style addictive algorithms.

Startups Outcomes

Speaking of Humane’s exit, some people say “9 of 10 startups fail”. That’s hard to verify, but how many startups actually make it to a successful outcome? Well, Carta puts out some data based on companies on their platform, which can give insight into the range of outcomes for VC backed startups.

Speed, Small Teams, Big Value

$100,000 salary used to be shorthand for “making it.” It’s still well above the median U.S. salary of $66,622, but in high-cost cities like NYC or SF, it no longer feels like the promised land of wealth. A similar shift is happening—much more rapidly—in startups.

$1M ARR is impressive. It signals a startup has moved from 0 to 1, has customers, and a potential growth path. But the perspective changing thanks (or no thanks) to AI companies. An investor I spoke with this week made two key points:

1. We’re in a paradigm shift—$10M ARR in the first 1-2 years is becoming common. They’ve seen more of these in the past six months than in the past decade. The gap between the best and the rest is widening.

2. The timeline to unicorn status is shrinking. It used to take around seven years. Now, if you don’t hit it in 2-3 years, you likely never will.

That’s just one investor’s perspective, but it reflects a broader acceleration, largely driven by AI startups.

VC firms need outsized winners to return their increasingly massive funds, which puts even more pressure on startups to grow at breakneck speed. In the worst cases, this could mean successful but slower-growing companies get written off. Ben Lang compiled a list of some of the fastest growing small teams, and their growth is impressive.

SIGNALS

• Microsoft unveiled a new quantum computing chip, “1/100th of a millimeter, meaning we now have a clear path to a million-qubit processor.”

• Mira Murati, OpenAI’s former CTO, is founding a new AI startup, Thinking Machines Lab. Murati (CEO) is joined by OpenAI veterans Barret Zoph (CTO) and John Schulman (Chief Scientist)

The Job Board

This week, a more unusual position in venture: A PhD fellowship at Positron Ventures. Traditional venture paths might look like Founder-Operator-Business School-Investor, but the venture market continues to cover all industries, and there are always roles for very smart people with niche expertise, especially in highly technical areas or the hard sciences. Job Below:

Positron Ventures

At Positron we believe that breakthroughs in physics, chemistry, and biology drive prosperity and are critical to solving global challenges in planetary and human health. Our mission is to proactively partner with scientists to bring more breakthrough innovations to market through startups. Our str

positron.ventures/news/positron-phd-fellowship

A word from a sponsor:

Accomplish More. Juggle Less.

When you love what you do, it can be easy to take on more — more tasks, more deadlines, more hours – but before you know it, you don’t have time to do what you loved in the beginning. Don’t just do more – do more of what you do best.

BELAY’s flexible staffing solutions leverage industry experience with AI systems to increase productivity without sacrificing quality. You can accomplish more and juggle less with our exceptional U.S.-based Virtual Assistants, Accounting Professionals, and Marketing Assistants. Learn how with our free ebook, Delegate to Elevate, and leave the more to BELAY.

Download now

On Inflation?

Happy Thursday. Just after we brought out the “mission accomplished” banner on inflation, the U.S. Consumer Price Index rose an annual 3% in January, likely delaying any expected Fed rate cuts. On top of this, Chevron and JPMorgan both announced layoffs (cutting over 9,000 jobs), which comes on the heels of Meta’s layoffs.

What’s that mean for venture: Capital is likely to remain expensive, which might depress VC fundraising. Startups should prepare for extended funding timelines, focus on profitability earlier…if they were not already.

While economic uncertainty can slow VC funding, downturns have produced some of the best startups (like Airbnb, 2008) as top talent enters the market. With skilled workers seeking new opportunities, we might see more startups, and startups will have a chance to hire experienced talent.

Community Spotlight + Toolbox

One of my favorite things about sending out Pitch Club is seeing members share what they're building, along with helpful tips and tools.

Pitch Club member and Techstars founder Sukhpal Saini is building Engyne.ai, a platform that helps you turn your knowledge into influence on LinkedIn. While working on Engyne, Sukhpal found himself creating a lot of LinkedIn carousels (those scrollable posts), which aren’t the easiest to design. So, he built a Canva tool to simplify the process:

Carousel Studio lets you create LinkedIn carousels directly in Canva in <5 mins with AI. 

How to use

• Pick a topic

• Click “Generate” and it will create a 4-slide carousel perfect for Linkedin for you

• Just edit & post!

It’s free, so if you are building a personal brand on Linkedin or want to play around on Canva, I highly recommend.

SIGNALS

(sorry, Musk is all over my feed these days)

• Musk’s X paid Trump $10M to settle his banning lawsuit….which seems, friendly.

• Musk also gathered funds to submit a $97.4B bid for control of the OpenAI non-profit board. Sama rebuffed the offer, which makes sense since OpenAI is reportedly in talks to raise $40 billion at a value of $260-300 billion.

• Founders Fund is reported to have raised $3 billion for their third growth fund.

• OpenAI released their product map, aiming to simplify their model offering and indicated GPT5 is “week/months” away.

Jobs in Venture

• Chief of Staff: Volt Capital, NYC

  • (Send a note to jane@volt.capital or soona@volt.capital as to why you’re uniquely suited for the role and please attach your resume.)

• Impact Investment Analyst: PureFuel, NYC (link)

Free Notion and Unlimited AI

Thousands of startups use Notion as a connected workspace to create and share docs, take notes, manage projects, and organize knowledge—all in one place. We’re offering 3 months of new Plus plans + unlimited AI (worth up to $3,000)!To redeem the Notion for Startups offer:

1. Submit an application using our custom link and select Beehiiv on the partner list.

2. Include our partner key, STARTUP4110P67801.

Apply Here

Forwarded this email? Subscribe here.

Happy Thursday.

Lunar New Year kicked off last Wednesday, bringing 15 days of celebration and the arrival of the Year of the Snake. For those celebrating, hope it’s filled with food, loved ones, and joy. 🐍

Signals

• Seed Round: Tandem, a marketplace for office space (and one of my favorite early startups), raised 6.1M in seed funding.

• Job Cuts: Workday announced layoffs of about 1,750 employees (8.5% of their workforce). Might be a omen of what’s to come in the jobs report from the Labor Department, releasing Friday.

• Internet Searching Outsourced: OpenAI launched Deep Research, an AI agent to do research online for you. Sounds useful (especially for newsletter writing), but it’s only available to the $200/month Pro users, so I guess I’ll stick with my 50 open tabs.

• In Case You Missed It: Daniel Penny, who was acquitted of choking to death Jordan Neely on a New York Subway, has been hired by Andreessen Horowitz. Mr. Penny was previously an architecture student and will be working as part of the “American Dynamism” team.

Field Notes

Apple unveiled Apple Invites on Tuesday, a new app for sending digital invites, managing RSVPs, and making party planning easier. And yes, it looks an awful lot like Partiful. Partiful, as one can imagine, is less than happy. In response to the launch, Partiful tweeted Apple’s own developer guidelines, which include the line:

"Come up with your own ideas… Don’t simply copy the latest popular app."

Image from Apple

Will Apple Kill Partiful?

As with many things Apple, the biggest advantage—and disadvantage—of Apple Invites is that it works best with other Apple people. Maybe this is an attempt to lock-in users to the ecosystem, back-stepping after they finally adopted RCS messaging in 2024 to improve texting with Android users.

Invites are tied to iCloud profiles. If you don’t have one, you’ll face extra steps like verification codes and limited access to in-app photo albums….sorry Android users.

Still it’s not great news for Partiful, who raised a 20M Series A in 2022, and would be likely to raise a larger round in the near, near future. They can only hope they have built a loyal enough base of users, and that Apple Invites ends up like Freeform, another pre-installed Apple app that quietly faded into the background.

The Age of the No-Code App?

Replit launched a version of their AI coding agent last year, but now it’s available for everyone to try—and it’s pretty killer.

Twitter tweet

For anyone who's ever wanted to build an app but didn’t want to deal with endless Stack Overflow searches (or doesn’t even know what that is), this could be for you. AI is making the idea of "no-code" more real than ever.

Even though we’re not at full "AI replaces engineers" mode yet, it is clear that there is traction. Cursor (AI Developer) just raised a $100M Series B—only four months after their Series A. $100 MILLION?!!! In 4 MONTHS!!?! That would sound absurd, but their parent company, Anysphere, is reportedly at $100M ARR, so in this AI economy that almost feels like a steal of an investment, and yes – people are using these tools and finding serious value.

Below is a great thread on the different AI coding agents available today. Worth checking out.

Twitter tweet

Jobs in Venture

• 📍Associate, Lerer Hippeau (New York)

• 📍Associate, Battery Ventures (San Francisco)

For Founders

• A reminder the Spring 25 Y Combinator Application is due Feb 11th at 8PM PT. The batch will take place from April to June in San Francisco.

  • https://www.ycombinator.com/apply

And a word from a sponsor:

Learn AI in 5 minutes a day

This is the easiest way for a busy person wanting to learn AI in as little time as possible:

1. Sign up for The Rundown AI newsletter

2. They send you 5-minute email updates on the latest AI news and how to use it

3. You learn how to become 2x more productive by leveraging AI

Sign up to start learning.

Thank you for reading. Forwarded this email? Subscribe Below:

The Company

Nguyen Coffee Supply, founded in 2018 by first-generation entrepreneur Sahra Nguyen, is America’s first specialty Vietnamese coffee company and has built a strong brand championing robusta bean coffee. They import through direct-trade relationships with Vietnamese farmers and roast in Brooklyn, New York. The company is on a mission to change the future of coffee through diversity, sustainability, and cultural integrity.

Product

Ground Coffee and Ready to Drink SKUs

Nguyen Coffee Supply offers ground coffee and ready-to-drink (RTD) beverages, focusing on robusta beans. The company is betting that robusta can become a powerhouse in the coffee industry, given its 60% less sugar, twice the caffeine, and higher antioxidants compared to arabica.

They launched DTC with whole beans in 2018, expanded into retail in 2021, and introduced their RTD line (Black, Coconut, Condensed Milk) in 2022, which fueled major growth. They also sell specialty brewing tools, including the phin, a traditional Vietnamese coffee filter.

Current State and Traction

Nguyen Coffee Supply has seen significant growth over the last two years, expanding its retail presence and proving demand for robusta-based coffee. While full financials aren’t public, here’s a look at their momentum:

• Retail Presence: Now in 2,500+ stores, including Whole Foods, Target, H-E-B, Sprouts, Safeway, and Vons.

• Wholesale Growth: Up 1,370% over the last two years.

• Retail Expansion: Grew its footprint 640% YoY, reaching 9,600 Points of Distribution (POD) in 2,000 new doors, including Target, Albertsons, Kroger, Sprouts, and H-E-B.

Whole Foods Performance

• Top 10 RTD coffee brand at Whole Foods and the only independent, non-VC-backed brand in the category.

• The Condensed Milk SKU has outperformed every Chamberlain Coffee RTD SKU in a 52-week period.

• Among the fastest-growing RTD coffee brands at Whole Foods nationwide.

Additional Highlights

• Winner of BevNET’s Best of 2024 Rising Star Award.

• RTD cost of goods (COGS) reduced by 37% in Q3 2024, improving margins.

What They’re Raising

Nguyen Coffee Supply has opened a $1.5M convertible note to fund their next big retail expansion.

• $18M pre-money valuation cap

• 10% discount and 5% interest rate

• Status: $300K committed

• Minimum Check: 25k

• Deadline: February 20, 2025

PitchClub Thoughts

Rhian and I angel invested in Nguyen in 2022 and in this round. Here are some of the things we assessed.

Founder

We were looking for a trusted, talented founder with founder market fit. Sahra had this and more, and is one of those founders who seems to have more hours in the day than the rest of us. She’s scaled Nguyen year over year, is careful about not raising more capital than needed, and has turned the company into a growth powerhouse with just a team of 5. Simultaneously, she’s the face of the robusta coffee movement. PR and media isn’t always a proper indicator, but based on our own interaction with Sahra, investor updates, referrals, and results, she’s as strong a founder as they come.

Interview with Sahra Nguyen
Sahra recently sat down with Whole Foods Market CEO, discussing the growth of Vietnamese coffee and the future of the industry. Watch here: YouTube Interview

Trends

The U.S. ethnic food and beverage market is rapidly expanding, with the Asian category expected to hit $98.06 billion by 2028, nearly doubling from $49.27 billion in 2021. Likely driven by the fact that the US Asian population is the fastest-growing demographic in America.

This has led to a boom of strong CPG brands focused on this shift, and we have seen strong fundraising momentum in adjacent categories:

• Xiao Chi Jie, a frozen dumpling brand, raised $10M in Series A

• Sanzo, an Asian-inspired sparkling water brand, also raised $10M in Series A, further validating consumer appetite for new beverage categories.

Exits

We, of course, assessed what a potential exit or positive outcome might look like for a specialty coffee brand. We felt The coffee industry had a strong history of high-value acquisitions, especially for brands that carve out a unique niche in the market, such as Nestlé acquiring a 68% stake in Blue Bottle Coffee for $700M in 2017 or Keurig Dr Pepper invested $300M for a 33% stake in La Colombe Coffee Roasters in 2023.

We saw exit potential in full acquisition, strategic investment from a large CPG player, or an international growth play.

Investment Type

CPG is HARD. We could (and probably should) write a whole newsletter edition on how brutal the CPG market has been the past 5 years. CPG businesses are capital-intensive, can be low-margin, and are often highly competitive. Especially recently:

• The cost of acquiring customers has grown

• Retail expansion can come with huge cash flow challenges

• CPG venture market has cooled

We feel Nguyen will be able to scale smartly, ride the growing demographic trends, and balance retail expansion with strong unit economics, but you might come to a different evaluation.

Sustainability

Rhian also was interested in the sustainability play of Nguyen. Robusta generally is naturally more resilient to climate change, requiring less water and pesticides, and grows at lower altitudes—all factors that could make them a more sustainable option as global coffee production faces climate challenges.

Connect

If you’re an accredited U.S. investor, please respond to this email, or reach out to me on Linkedin, for the full pitch deck and an introduction to Sahra Nguyen.

Submit your Startup:

If you run or work at, or have invested in an early stage startup that’s looking for early adopters, investors, or just want to get the word out there, please share your story with us, and we would be excited to partner with you.

Submit Your Startup

Forwarded this email? Subscribe Below:

Good Morning! Welcome to the Thursday update. We’ve got consolidation, distillation, and jobs today.

Heat Map: Rich VC Firms get Richer in 2024

2024 was a tough year to raise VC capital—unless you were already at the top. According to PitchBook, half of all venture capital raised this year went to just nine firms. The driving factors: AI, vertically integrated VC funds, and access to massive LPs like sovereign wealth funds, according to VC reporter Rosie Bradbury.

While boutique and emerging firms still pulled in over $9 billion, the consolidation of capital at the top has undoubtedly changed the ecosystem. Smaller firms now struggle to access competitive deals, participate in mega-rounds fueling the AI arms race, and secure LP capital. This shift may push smaller and solo-GP funds toward more niche areas and investment theses, finding value where the big players overlook.

Distilling Models: Smarter, Smaller, and Maybe Stolen?

DeepSeek followed the release of their R1 model with an image model, Janus-Pro-7B, but skepticism around DeepSeek's training claims escalated to full accusations of model distillation.

What is Model Distillation?

Model distillation is the process of taking a large AI model and creating a smaller, more efficient version that retains most of its intelligence. Instead of training from scratch on raw data, the smaller (student) model learns directly from the larger (teacher) model’s outputs — or uses a large model to create probabilistic distributions for a classification. So instead of training on books and random Reddit threads, the distilled model gets to train on pairs of prompts and completions or ready-made probabilities of “correct” completions, which is way better.

The big model has already done the heavy lifting—understanding complex tasks, reasoning, and making decisions — the distilled model just needs to mimic that knowledge efficiently, and can often do so with smaller parameters. After training, it can generate nearly the same outputs as the teacher model but requires significantly fewer computational resources.

OpenAI does this internally for practical reasons: smaller models run faster, use less power, and cost less to deploy, like them distilling GPT-4 to GPT-4-turbo. OpenAI even offers it as a service. But if another company distills an AI model trained by OpenAI—especially without permission—it can look a lot like stealing.

DeepSeek Caught Red-Handed?

It’s hard to say this definitively. OpenAI says they found suspicious activity and evidence that DeepSeek is using its models in this way. If DeepSeek DID train on OpenAI’s outputs, this would be a textbook example of "model extraction". It will be hard to prove, but there might be signs, like OpenAI model quirks appearing in DeepSeek’s responses, or R1 thinking it is made by OpenAI (which people are reportedly seeing).

Frontier and Safety Concerns

We often hear about AI companies moat (or lack thereof). One of the biggest moats in AI is not just the raw model but its thinking process. Advanced techniques like Chain of Thought (CoT) reasoning allow models to break down complex problems, improving accuracy and reducing hallucinations.

If another company manages to extract CoT reasoning, this could significantly narrow the performance gap between AI leaders and challengers. It also raises concerns about safety, as smaller, unregulated AI models might lack the alignment, security, and deployment measures agreed to by frontier labs.

Is This Illegal?

Right now, it’s a legal and ethical grey area, where the law hasn’t really caught up (people are comparing it to OpenAI’s own alleged training on copyrighted material). It’s certainly against OpenAI’s terms of service. OpenAI disallows this practice because it works — it’s an effective way to shortcut AI development. If regulators don’t step in, expect more companies to attempt the same strategy.

Woof, ok, that’s enough AI and DeepSeek talk for this week. But also, do you think DeepSeek used distillation for its model?

Toolbox

GP Ryan Hoover successfully transitioned from founding Product Hunt to running a thriving venture fund. Here’s an example of a successful pitch deck from Weekend Fund’s third fundraise. View the Deck:

Weekend Fund Deck

Thanks for reading. We hope everyone has a wonderful weekend.

A short message from a sponsor:

Get your news where Silicon Valley gets its news 📰

The best investors need the information that matters, fast.

That’s why a lot of them (including investors from a16z, Bessemer, Founders Fund, and Sequoia) trust this free newsletter.

It’s a five minute-read every morning, and it gives readers the information they need ASAP so they can spend less time scrolling and more time doing.

Subscribe for free

Welcome to the first Pitch Club of 2025. After a short break, we’re back and ready to dive into an exciting year ahead.

Field Notes

This week, Chinese AI startup DeepSeek launched their latest model, DeepSeek-R1, claiming it rivals one of OpenAI's top models, o1. It matters because (they say) it was built for ~$6 million, which is insanely cheap (Gemini cost ~$191 million). DeepSeek reportedly trained the model on fewer of Nvidia’s high-end chips than OpenAI, sending Nvidia’s stock into a tailspin—dropping 17% and wiping out $600 billion in market value, sending shockwaves through the industry. For startups, R1 has been met with a mixed reaction, depending on where you’re building:

• Model Layer: For companies and investors who’ve poured billions into building SOTA models, this is nothing but bad news. It raises a tough question: What happens when a smaller player leverages the costly legwork and catches up anyway? For industry darlings, it’s a reminder that even the biggest lead can vanish quickly.

• Application Layer: For startups in this category, this could be a game changer. Open weights and cheaper models mean faster, more affordable access to cutting-edge AI tools—making your tech smarter without the giant budget.

Some technologists and investors remain skeptical of DeepSeek's claims about model training, but if they hold up, it could signal a new paradigm for the already volatile AI startup ecosystem—and a huge boost to those betting on open-source AI as the ultimate winner.

Signals

• Origami Agents, a YC backed startup building AI research agents for sales and GTM raised a $2M seed round.

• Barclays is closing Rise, the NYC fintech accelerator that helped launch Alloy, Chainalysis, and Novo Bank.

• Osuke Honda, Ken Hara, and David Cheng launched Coreline Ventures focusing on early-stage entrepreneurs in the US, Japan, and Korea.

Submit your Startup:

Speaking of the upcoming year, if you run or work at, or have invested in an early stage startup that’s looking for early adopters, investors, or just want to get the word out there, please share your story with us, and we would be excited to partner with you.

Got Signal?

Send any startup tips, news, suggestions by replying to the newsletter. We hope everyone has a great start to the week!