Data to deployment
1. Key Themes
Theme 1: Value in the data market accrues by moving up the chain, from raw data toward deployment
The raw-data seller is stuck at the bottom of the stack
Raw corpus is the least differentiated, least valuable layer, and frontier lab buying behavior makes it worse.
"So the worst place to be is the bottom of the totem pole, dealing CSVs with little to no value added beyond origination. The frontier labs' buying behavior exacerbates this. The big labs only want to buy from a small number of vendors at a time."
The spread between layers makes the strategy obvious
"If the spread between a raw corpus and the tasks you build on it is 2 to 5x, the obvious thing to do is not to get raw data and sell it, but to get raw data, build tasks, and sell the tasks. The same is true from tasks to environments, environments to post-training, and post-training to deployment."
Data vendors are already converging on becoming deployment businesses
"The most durably valuable companies in this economy will keep moving downstream, becoming more full stack deployment businesses... The emerging shape of these businesses seems to be converging on a combination of the data pipeline, deployment and integration, and eventually inference."
Theme 2: Today's data market is immature and will commoditize and financialize
No units, no prices, wide spreads
"Looked at like a financial market the dislocations are clear: there's no standardized pricing, there's no standardized unit of value, the spreads are very wide, the transaction costs are very high."
Private data is the next frontier, and it's currently valued at zero
"The world's public data is now incorporated into the models; further improvement will rely on novel methods and novel (private) data. That private data is sitting completely unaccounted for on company balance sheets and in people's heads."
Expect a quasi-commodity market, not "the new oil"
"So data will become a quasi-commodity... it will look like more of a commodity market than it does right now: more efficient, more standardized, more liquid, without ever being perfectly efficient, standardized, or liquid."
Theme 3: The market will bifurcate into synthetic volume and scarce novelty
The "Walmart of tokens" vs. the novelty market
"High volume, low value on a unit basis. This will likely approach 100% synthetic and simulated: the Walmart of tokens. Low volume, high value. Here the value comes not from scale but from novelty: the truly out-of-distribution data and the novel signals that emerge from the real world usage, and the edge cases."
The two halves reward different skills
"The real winners in the high volume, synthetic data market will look like extremely talent dense research companies rather than the hustle-y dealmakers in the real world data business... It will look and feel like market makers and quant shops. Contrast that to the out of distribution data, which will continue to be sought after and valued more like VC/PE deals."
The split could happen fast
"Given the speed at which things are moving with the labs and data business, this split could happen quite quickly (within the next year)."
Theme 4: The firm becomes a "Task Factory," and novelty never runs out
Deploying AI generates new edge cases by definition
"This is the vision for the firm as a Task Factory: as you deploy AI into a business, you are inventing new and novel failure cases for the models, because you're bringing them up against real-world situations."
Simulation and reality are complements
"Pure simulation won't give you the edge cases. Real world experience won't give scale."
The machine economy produces fundamentally new data
"As the substrate of the economy moves from human hours to tokens and GPU cycles, the machine economy will generate data and outcomes that is not merely better or worse than the human economy's, but wholly different through the very fact of its being AI produced."
Theme 5: Data enters the capital structure, and data allocation becomes a core CEO decision
Data is a $0 intangible today, which creates arbitrage
"Today, data is sitting unaccounted for on balance sheets as a $0 intangible... The gap between what data sells for and what it's carried at is a ton of near-term arbs and, longer term, room for new capital structures."
AI rollups and data-native LBOs are the vehicle
"The vehicle that puts it together will be AI rollups with data as a core part of the buying criteria and value creation strategy... Fully financed data - the static corpus and the new potential of the task factory - will be the new junk bond that will spur the new LBO."
Data allocation parallels capital allocation
"Making the choice between keeping data internal vs. exposing it externally (and how) is akin to the choice between returning capital to shareholders and letting it compound (and how)."
2. Contrarian Perspectives
The bear case is a critique of today's market, not of the durable market
Rechtman concedes the bear case (non-rivalrous data, falling prices, synthetic substitution) but argues it doesn't kill the long-term market.
"Ultimately the bear case is a repudiation of the current market rather than an indictment of the long term viability of the data market. It is obvious to anyone paying attention that the market regime today is ridiculous, unsophisticated, and not durable - and it does not deserve to last."
His durable demand case rests on owned intelligence and ROI-based allocation: "You don't need frontier intelligence for most things, so most of a company's daily work gets allocated like labor, on ROI, where post-trained models and/or harness + eval win."
Data is NOT "the new oil"
Against a popular analogy, he gives two structural reasons.
"First, it's non-rivalrous. There's nothing inherent to a CSV that prevents both of us from owning it... Second, data and its byproducts are too highly varied to clear in a single market the way oil or compute do. Raw data is a different thing than tasks, and different tasks are different from other tasks or environments."
AI diffusion will be much slower than Silicon Valley assumes
"So AI diffusion is going to be much slower than the SF groupchat thinks. We're 25 years into the cloud and it's not done yet. Migrating to the cloud meant rearchitecting technology systems, which was relatively simple. AI is rearchitecting the economy."
He also argues that full internalization of training data by labs is "years away" because new domains will be "further from code and less verifiable."
3. Companies Identified
Waymo
- Description: Autonomous driving company.
- Why mentioned: Historical precedent for the "simulation seeded by real-world edge cases" model that he expects to repeat in a mid-diffusion economy.
- Quote: "Waymo is the historical precedent; it has a huge amount of simulated experience but originates the things it needs to simulate from real-world context. There's some weird edge case Waymo may only see once... But Waymo sees it, it can simulate it a billion times and figure out how to solve it. That one real-world instance is novel, valuable, and out of distribution."
- Description: AI rollup company (linked in the article).
- Why mentioned: Named as the early leader of AI rollups that treat data as a core buying criterion and value-creation lever.
- Quote: "The vehicle that puts it together will be AI rollups with data as a core part of the buying criteria and value creation strategy (CAC is the early leader)."
Frontier labs (unnamed)
- Description: The large AI labs that are the primary data buyers.
- Why mentioned: They concentrate demand and shape vendor economics, spending up to nine figures a month.
- Quote: "It's uneconomic to manage infinite vendor relationships at the scale they're deploying capital (as much as 9 figures per month for some labs)."
Slow Ventures
- Description: Venture firm where the author is a partner.
- Why mentioned: The author's employer; he leads pre-seed/seed rounds from a ≈$325M fund.
- Quote: "I'm a partner at Slow Ventures, where I lead pre/seed rounds from a ≈$325M fund."
4. People Identified
- Description: Partner at Slow Ventures and author of 99D.
- Why mentioned: Author of the piece, drawing on 60+ calls across the data supply chain.
- Quote: "Over the last month I've done 60+ calls with people up and down the data supply chain: brokers, wholesalers, researchers, frontier lab buyers, RL and data labs leaders, etc."
No other individuals are named in the article.
5. Operating Insights
Don't be a pure broker: you need proprietary origination or value-added services
"Without some special way to buy (proprietary origination) or sell (value added services) then you're competing on price and execution. If you have no reason to think there's gonna be long term margins then that's just brokerage. For most businesses, being full-time brokers is a mistake. Lab needs change fast enough that the air goes out of any one thing."
Diversify away from the small set of frontier lab customers by moving into enterprise deployment
Lab concentration is a business risk, and moving downstream creates more terminal value.
"This is necessary to diversify away from the limited universe of frontier lab customers and to create something with more terminal value - the next gen management and IT services firms."
Be aggressive now, while the market is dislocated
"If you can be really aggressive, you can probably hoover up and create a lot of value for yourself while we're in this highly dislocated market."
Concrete plays he highlights include passive extraction at scale ("find thousands of suppliers, get the permission, structure the data, and get suppliers paid on consumption instead of one corpus sale at a time") and building the "token X-ray" to value datasets prospectively.
6. Overlooked Insights
Why labs buy through few intermediaries: provenance liability and signaling
Beyond economics, the vendor-concentration pattern is partly a risk and secrecy mechanism, which implies intermediaries have a structural role.
"Transacting directly with many vendors exposes them to increased liability arising from the data provenance... It's impossible to communicate to the market without exposing research priorities and product roadmaps. Working through a limited number of intermediaries helps obfuscate their buying behavior similarly to financial markets where institutions might buy via a dark pool or prime broker offering swaps."
Robotics data is about to repeat the business-data playbook, imminently
A near-term timing call that is easy to skim past.
"It's very clear that what's currently happening in business data will come for robotics data within the next few months and reach a fever pitch over the next few years. The integration ➡️ transformation ➡️ origination playbook will look different in robotics data (it's more origination than extraction) but the fundamental principle is the same and is very promising."