Sarah Guo Is Betting Nearly a… | The AI Corner Summary

1. Key Themes

Theme 1: The Application Layer Above Foundation Models Is Where the Real Returns Live

Guo's entire investment thesis rests on the belief that foundation models function as infrastructure — powerful but ultimately a metered input — while the durable value accrues to companies built on top of them.

"Sarah Guo built Conviction around a simple thesis: the biggest AI opportunities lie above the foundation models, not inside them."

"Inference is sold by the token the way power is sold by the kilowatt-hour, which makes a lab a metered input to other people's businesses rather than a competitor for end-user preference."

Theme 2: AI Is Unlocking Labor Markets, Not Just Software Markets

The most significant market expansion isn't winning budget from existing SaaS buyers — it's converting labor spend (legal, clinical, educational) into software revenue. This is a structurally larger and less contested opportunity.

"Legal is a services market rather than a software market, and doing the low-level work is a bigger pie than selling tools to the people who currently do it."

"Her counter to the consolidation worry is narrow. There are far more markets now addressable by software, because willingness to pay is being drawn from budgets that were never software budgets. Legal work, clinical judgment and education were all priced as labour."

Theme 3: Organizational Physics Limits What Labs Can Build

Large organizations have one A-team. Everything else is under-resourced. This structural reality means labs will remain focused on model development and cannot competently colonize the application layer.

"Any large company has exactly one A-team, and the vast majority of its product surface is not staffed by it. If your product is Google's 27th priority, you are not competing with Google. You are competing with a funded but unloved product group."

"Her claim about applications is carefully worded, saying that the labs have tried to build horizontal and vertical products and have not yet succeeded, which in her view is not because they fail to see the enterprise value, but rather because knowing what customers want, winning distribution, and rebuilding continuously around capability that is not your own are all genuinely difficult."

Theme 4: Concentration Risk in Venture Capital Is Historically Unprecedented

Two-thirds of all venture dollars in Q1 2026 flowed to just four companies. This level of concentration reshapes the landscape for early-stage investors and creates structural pressure on the thesis that startups can win.

"Anthropic, OpenAI, xAI and Waymo absorbed roughly 65 cents of every venture dollar deployed in the first quarter, out of a record $300 billion. Anthropic alone went from a $9 billion revenue pace in January to a $47 billion run rate five months later."

"Concentration on this scale has no real precedent in venture capital, and it makes the size of Guo's operation look like a rounding error."

Theme 5: Open-Weight Models Are Creating Structural Leverage for Startups Against Labs

When frontier lab pricing spikes, open-weight alternatives from China and Europe serve as credible pressure valves — strengthening the hand of application-layer companies that can route around proprietary APIs.

"Token prices rose this year, in some cases by a factor of 100, and enterprise buyers began objecting in public. Open-weight models out of China and Europe now land within reach of the frontier at a fraction of the cost, giving those buyers a credible alternative."

"One of her founders said he had 'no intention of spending his career drinking Anthropic and OpenAI's water.'"

2. Contrarian Perspectives

Perspective 1: More AI Winners Doesn't Mean More Software Companies — It Means More Markets

The consensus fear is that AI consolidates into a few giants, squeezing venture returns. Guo reframes this entirely: the number of markets addressable by software is expanding, not contracting, because AI is converting labor into software. This is a different argument than "startups will beat the labs."

"There should not be as many SaaS companies as there are. She volunteered that venture has always been an outlier business and is becoming more of one."

"There are far more markets now addressable by software, because willingness to pay is being drawn from budgets that were never software budgets."

Evidence: Harvey — valued at $11 billion, $300M revenue, 100,000+ lawyer users — is extracting revenue from the legal services market, not the legal software market. That's a fundamentally larger TAM.

Perspective 2: As Models Improve, the Human Work Around Them Gets Larger, Not Smaller

Conventional wisdom assumes better AI = fewer humans needed. Guo's observation from Harvey inverts this: trust delivery still requires people, and the complexity of helping humans extract value from AI scales with model capability.

"Harvey, the company built to do the work of lawyers, employs 200 lawyers of its own, many of whom teach other lawyers in person how to let the software do their jobs. What it actually sells beneath the product is trust, and trust still gets delivered by people."

"As the models improve, the work of helping a human extract value from them gets larger rather than smaller."

Evidence: Harvey employs 200 lawyers internally despite being an AI-first company, at $11B valuation and $300M revenue — suggesting the human-in-the-loop layer is a feature, not a bug.

Perspective 3: Early-Stage Venture May Now Be a Sourcing Funnel for Late-Stage Platforms

Guo herself names the most threatening structural risk to her own strategy: that seed-stage venture is becoming a subsidized discovery mechanism for larger players with cheaper capital — not an independent value-creation engine.

"Early-stage venture may now function as a sourcing funnel for late-stage platforms with a lower cost of capital who can subsidise competitiveness at seed out of growth fees."

"Her rebuttal, that she needs only a couple of important companies per cycle, is an assertion about her own selection ability rather than an answer to the structural point."

Evidence: OpenAI seeded Harvey, then built its own legal tools. It led the seed round in Cursor, then competed directly in coding. OpenAI also acquired Remotion from Guo's portfolio. The labs are demonstrably using the ecosystem as a product roadmap.

3. Companies Identified

Conviction Description: Early-stage AI-focused venture firm founded by Sarah Guo in October 2022. Why Mentioned: Central subject of the article; the firm whose thesis and performance are being analyzed.

"Conviction employs 8 people. 4 of them invest… There are 3 funds now, totalling close to a billion dollars."

Baseten Description: ML model deployment infrastructure company; allows companies to run their own models on their own terms. Why Mentioned: Conviction's highest-profile portfolio success; proof-of-concept for the thesis that open-weight model adoption would spike when lab pricing rose.

"Revenue grew 20x in 12 months, inference volume grew 40x, and the company repriced from $5 billion in January to $13 billion five months later."

Harvey Description: AI-native legal services company; applies AI to do actual legal work, not just sell tools to lawyers. Why Mentioned: Prime example of Guo's "labor market, not software market" thesis; also illustrates the counterintuitive human trust layer.

"The company is valued at $11 billion, revenue tripled in the past year to $300 million, and more than 100,000 lawyers run work through it. Guo wrote the first cheque personally, before Conviction had a fund, to two founders on a bad Zoom call with no slides and no prototype."

Anthropic Description: Frontier AI lab; one of the dominant foundation model providers. Why Mentioned: Used as the clearest example of a lab whose A-team is entirely consumed by model development, validating the organizational physics argument.

"Anthropic's A-team is the model itself, with a thesis centred on code and self-improving systems, and she treats the decision not to build video and image models as evidence of focus rather than only of safety policy."

OpenAI Description: Frontier AI lab and Conviction's most direct strategic threat. Why Mentioned: Named as the clearest evidence that the thesis has "already taken damage" — it seeded Harvey, led Cursor's seed, then competed directly in both categories.

"That assumption has already taken damage when OpenAI seeded Harvey in 2022 and led the seed round in Cursor, back when the labs still funded the ecosystem above them. Both now sell legal tools of their own."

Casa Systems Description: Broadband equipment company founded by Jerry Guo; went public at $1.2B market cap in 2017 before filing Chapter 11. Why Mentioned: Origin story for Sarah Guo's competitive instincts and tolerance for existential risk — her formative professional environment.

"A large incumbent sued Casa on what turned out to be no real legal grounds, the case settled for nothing, and the legal fees that year exceeded the entire revenue of the business."

Thinking Machines (Mira Murati's company) Description: AI company founded by former OpenAI CTO Mira Murati. Why Mentioned: Briefly mentioned as a neighbor to Conviction's office; signals the density of AI activity in the Mission District.

"The office sits on York Street in the Mission District of San Francisco, a few blocks from the building Elon Musk leases for xAI and a short walk from Mira Murati's Thinking Machines."

4. People Identified

Sarah Guo Description: Founder of Conviction; former partner at Greylock; daughter of Casa Systems founder Jerry Guo. Why Mentioned: Central subject; the investor whose thesis, track record, and intellectual honesty the article profiles.

"She debuted at number 56 on the 2026 Forbes Midas List, and of the roughly 21 AI-native companies that have crossed $10 billion in valuation on revenue run rates above $100 million, Conviction has backed 6."

Mike Vernal Description: Partner at Conviction; previously a partner at Sequoia and one of Facebook's most senior product leaders. Why Mentioned: Key hire that helped scale Conviction alongside its second fund close.

"A second fund closed at $230 million alongside the hire of Mike Vernal, previously a partner at Sequoia and before that one of Facebook's most senior product leaders."

Jerry Guo Description: Sarah Guo's father; immigrant from Hunan, China; founder and former CEO of Casa Systems. Why Mentioned: His entrepreneurial arc — from $50 in 1987 to a Nasdaq IPO to bankruptcy — directly shaped Sarah's worldview on competitive resilience.

"Her father, Jerry Guo, arrived in the United States from Hunan in 1987 with $50 and a transcript from Tsinghua, having placed first in the country on the gaokao."

Andrej Karpathy Description: Prominent AI researcher; co-founder of OpenAI; formerly worked out of Conviction's office. Why Mentioned: His departure to join Anthropic is cited as a tangible sign of ecosystem talent flowing back toward labs — a threat vector to the application-layer thesis.

"Andrej Karpathy, who worked out of her office, joined Anthropic in May."

Mira Murati Description: Former CTO of OpenAI; founder of Thinking Machines. Why Mentioned: Brief geographic reference establishing the concentration of top AI talent near Conviction's office.

"A short walk from Mira Murati's Thinking Machines."

5. Operating Insights

Insight 1: Assume the Music Stops — Model It Before It Happens

Guo's standing advice to founders is to war-game what a market contraction or funding freeze does to the business before it happens — not as a theoretical exercise but as a planning tool.

"She tells founders to assume the music stops and model what it does to the business."

Tactical application: Build a "music stops" scenario into quarterly planning. What does your burn, revenue retention, and customer concentration look like if fundraising markets close or lab pricing doubles again overnight?

Insight 2: The Feature-to-Wedge-to-Platform Playbook Is No Longer a Reliable Map

Guo explicitly abandoned the classic SaaS progression framework after seeing it fail to predict success in the AI era. Founders still anchoring strategy to this model may be optimizing for the wrong signals.

"She has also abandoned the classic progression from feature to wedge to platform, which she now says correlates very little with success in either era."

Tactical application: Stop treating platform ambition as a proof point in fundraising. Investors like Guo are evaluating whether you're solving a real labor-priced problem with distribution advantages — not whether your roadmap ends in a platform.

Insight 3: Size Your Winners More Aggressively Early — Don't Let Constraint Become Theology

Guo's own stated regret is that early positions in breakout companies were too small. Her original view — that scarcity disciplines investors — hardened into a doctrine that cost her ownership in her best companies.

"Her stated regret is that the early positions were sized too cautiously… By 2026 there are three funds near a billion dollars, she has invested in every Baseten round with each cheque larger than the last, and she co-led a $1.5 billion Series F."

Tactical application: When a portfolio company shows clear signs of category definition (not just growth), resist reserve discipline and lean into follow-on. Intellectual consistency is not a virtue when the facts have changed.

6. Overlooked Insights

Insight 1: Raising Money Is Now Easier Than Making Money — and That's the Problem

Buried in a list of Guo's self-critical observations is a claim with significant implications for both founders and investors: the current environment has decoupled fundraising skill from value-creation skill in a way that will eventually correct.

"She has said that raising money is now considerably easier than making money, that the current expansion will not end well for returns, and that a founder arguably should not want their investor's multiple to be high, because it came out of their dilution."

This is a subtle but important point for founders: high valuations driven by competitive financing rounds are not neutral. They are a direct transfer of future equity value from founders to early investors seeking high multiples.

Insight 2: The Thesis Remains Unproven — No Portfolio Company Has Gone Public

The entire Conviction thesis — and its $1 billion in AUM — is still a paper argument. The article's most understated sentence may be its most important one for investors evaluating Guo's track record.

"Everything so far is prologue, because not one company in the portfolio has yet rung the bell on a public exchange. Until one of them does, the most rigorously argued position in venture capital is still a very expensive opinion held with unusual… conviction."

Harvey at $11B, Baseten at $13B — these are venture marks set by other VCs in a record deployment quarter, not realized returns. The thesis has yet to be tested by public market scrutiny.