Teahose.
SIGN IN
NEW HERE — WHAT TEAHOSE DOES
We read the entire AI & tech firehose — so you don't have to.
PODPodcastsAll-In, No Priors, Acquired…
NEWNewslettersStratechery, Newcomer…
PAPPapersPhysical AI research
PHProduct Huntdaily launches
VCInvestor ScoutSequoia, a16z, Benchmark…
CLAUDE DISTILLS →
7 reads, 30 sec each — free, 6 AM ET.
+ a live graph of the companies, people & themes underneath.
HOME/99D/Spiritual salvation through arti…
NEWS
// NEWSLETTER ISSUE
99D

Spiritual salvation through artificial intelligence

DATE November 20, 2026SOURCE 99DPARTICIPANTS YONI RECHTMAN
In this episode
// SUMMARY

1. Key Themes

Theme: Training-data, post-training, and inference companies are all converging on a single business model: AI transformation/deployment

The convergence is industry-wide, not limited to data companies

Rechtman argues the whole supply chain is trending toward one shape.

"Every company across the training data supply chain is converging on the same business model, which is essentially an AI transformation business. And it's not just the data companies; it's the post-training and inference businesses too!"

"Base 10 and Mercor and Applied Compute and the rest are all going to trend to basically the same company because there is only one business to build and one product to sell the enterprise: AI transformation."

Even coding-agent players are drifting this way

"Arguably, even Cognition is ending up in the same direction, as they really emphasize their deployment motion to differentiate their coding agent."

The logic: climb the value chain toward customers

"If you accept the claim that there's a better business as you get closer to customers, and you accept the obvious reality that ultra-high customer concentration is not good, then the natural conclusion is that people will just keep climbing up the value chain toward the ultimate end of offering AI transformation as a service writ large."


Theme: Commoditization of the lower layers; value accrues to delivering business outcomes

Data, models, and inference are commoditizing

"Everything below that is commoditizing. Data, models, inference, etc. The thing that isn't a commodity is delivering a business outcome. The pitch is just: I'll do AI for you."

Enterprises buy change, not software

"Nobody wants to buy software. Buying software sucks. People want to buy a change in their business, and you can charge a lot more and have an easier time selling it if you just say: I'm going to sell you the change you want to make."

Only two companies are left to build

"If you're not doing hardware or techbio or foundation models, the only software company left is selling AI transformation, whether as a product or as a service."

"There are really only two companies to build: you build intelligence or you diffuse it."


Theme: The data business is a funding and access vehicle, not the destination

High cash flow, low terminal value

"The alternative is staying in the data business and recognizing it's high cash flow, low terminal value. Otherwise you overfit the current moment. Building an RL environment might be a great thing to do right now. It's not the thing you'll be doing for the rest of your life."

The playbook: data first, enterprise second

"Build a data business like what Mercor and Scale and Handshake have actually done: get into this super fast-moving, high-dollar flow from the labs, and use that to raise a bunch of money and generate a bunch of cash in the data business while getting a lot of proximity to the labs and credibility in the eyes of the rest of the world. Then go and build an enterprise AI company."

Venture scale requires a second act

"Venture scale in data means finding your way to a second act funded by the data business."


2. Contrarian Perspectives

Contrarian view: Early-stage companies should do "anything" and follow momentum, rather than stick to a focused thesis

Rechtman admits this cuts against his normal investing instincts, and argues the size of the money flows rewards opportunism.

"This is anathema to what I normally believe as an investor. When I look at the most successful companies in this space, they basically started by doing anything and then followed the momentum."

"The money sloshing around is so big that it rewards you for backflipping your way to accumulating the catalytic materials to build an important business: proximity to the labs, proximity to enterprises, talent, capital. And you don't accumulate those things by sitting around like a monk. You do it by doing stuff."

Contrarian view: Enterprise AI has almost nothing to do with the training data business; the data business is just a credibility engine

The skills are not transferable. What transfers is trust and access.

"No, enterprise AI has almost nothing to do with the training data business, except that by running the training data business they've been able to get close with the labs and close with the enterprises. And the enterprises trust them because they're close with the labs. 'Hey, we do a lot of business with the labs' is a good way to have credibility in the eyes of, say, Bank of America."

Contrarian view: Application-layer and deployment companies are fundamentally the same business

"The reason a bank buys Decagon is the same reason it buys Base 10."


3. Companies Identified

Mercor

  • Description: AI training data/expert labor marketplace serving the frontier labs.
  • Why mentioned: Cited as validation of the thesis that data companies become enterprise AI deployment businesses, and as a model of the data-to-enterprise playbook.
  • Quotes: "Almost immediately afterwards, Mercor proved my point!" / "The Mercor chart itself is obviously about customer count, not revenue (which we know is concentrated at the labs), but it's telling nonetheless. Enterprise AI is the story about the future and how they see the business."

Base 10 (Baseten)

  • Description: Inference/model deployment infrastructure company.
  • Why mentioned: Example of a non-data company converging on the AI transformation model; also used as the comparison for why banks buy AI products.
  • Quotes: "Base 10 and Mercor and Applied Compute and the rest are all going to trend to basically the same company..." / "The reason a bank buys Decagon is the same reason it buys Base 10."

Applied Compute

  • Description: Post-training/enterprise AI company.
  • Why mentioned: Named as another company trending toward the same AI transformation model.
  • Quotes: "Base 10 and Mercor and Applied Compute and the rest are all going to trend to basically the same company..."

Scale AI

  • Description: Data labeling/training data provider.
  • Why mentioned: Example of a data business that used lab proximity and cash flow as a foundation.
  • Quotes: "Build a data business like what Mercor and Scale and Handshake have actually done..."

Handshake

  • Description: Training data/expert network business (expanded from its career-platform roots).
  • Why mentioned: Cited alongside Mercor and Scale as having executed the data-business playbook.
  • Quotes: "Build a data business like what Mercor and Scale and Handshake have actually done..."

Palo Alto (Networks)

  • Description: Cybersecurity company.
  • Why mentioned: Its deal messaging is cited as evidence for the convergence thesis (details not included in the text; likely shown in an image).
  • Quotes: "The messaging around Palo Alto's deal to makes this point clear."

Cognition

  • Description: Coding agent company.
  • Why mentioned: Arguably heading in the same direction by emphasizing deployment as a differentiator.
  • Quotes: "Arguably, even Cognition is ending up in the same direction, as they really emphasize their deployment motion to differentiate their coding agent."

Decagon

  • Description: AI customer support agent company.
  • Why mentioned: Used as an example of a product that banks buy for the same underlying reason they buy deployment services.
  • Quotes: "The reason a bank buys Decagon is the same reason it buys Base 10."

Bank of America

  • Description: Large financial institution.
  • Why mentioned: Illustrative enterprise customer whose trust is earned via lab proximity.
  • Quotes: "'Hey, we do a lot of business with the labs' is a good way to have credibility in the eyes of, say, Bank of America."

Slow Ventures

  • Description: Venture firm where the author is a partner, with a roughly $325M fund.
  • Why mentioned: Author's firm, which leads pre-seed/seed rounds.
  • Quotes: "I'm a partner at Slow Ventures, where I lead pre/seed rounds from a ≈$325M fund."

4. People Identified

Brendan Foody

  • Description: Mercor leader, source of the Mercor chart.
  • Why mentioned: His chart/post (about Mercor's customer count) is the evidence the author cites.
  • Quotes: "From Brendan Foody" / "Almost immediately afterwards, Mercor proved my point!"

Yoni Rechtman

  • Description: Author; partner at Slow Ventures leading pre-seed/seed rounds.
  • Why mentioned: Author of the piece, offering the thesis.
  • Quotes: "I'm a generalist investor looking for weird takes on important stories: N-of-1 companies taking non-obvious approaches to markets that matter."

5. Operating Insights

Use the data business's cash and credibility to fund a second act. Treat the current high-cash-flow business as a bridge, not a destination.

"So they're using all of the cash flow they generate from this low terminal value training data business to just invest in enterprise AI."

Sell the outcome, not the software. Positioning as a change in the customer's business is easier to sell and commands higher prices.

"People want to buy a change in their business, and you can charge a lot more and have an easier time selling it if you just say: I'm going to sell you the change you want to make."

Beware customer concentration and overfitting the current moment. Revenue concentrated at a few labs, and products tied to today's trends (like RL environments), are traps.

"Building an RL environment might be a great thing to do right now. It's not the thing you'll be doing for the rest of your life."


6. Overlooked Insights

Differentiation will come from specialization, not from the core offering. Rechtman expects a proliferation of deployment companies segmented by motion, product, service, or customer rather than a single winner.

"You can slice is N ways (different motions, products, services, customers, etc.), but that's going to be the only company that exists. Hundreds of deploycos with different specializations."

Deployment is the endgame for every enterprise business, not only AI vendors. The framing implies incumbents, too, will converge on AI transformation.

"Enterprise AI is the final act of all of their businesses because it's the final act of EVERY enterprise business now."