Who Buys a Dead Startup’s Data? 36 People in AI Pick micro1

Chris Barber asked 36 people in AI who to talk to if you are selling a private dataset, “like a shut-down startup’s data”, and micro1 drew 19 votes, according to results he posted on X on 29 October. Mercor was second with 9 votes, and Handshake and SF Data had 5 each.

By our arithmetic, micro1’s 19 votes were about 30% of the 64 votes listed across 19 companies, and more than Mercor’s 9 and Handshake’s 5 combined. Barber discloses that he is a small investor in several companies on the list, and he says he did not vote.

Short explainer · Private data buyers

A vertical news Short narrated from this piece’s headline and opening line.

Business Pill · Data Labeling

An evergreen Business Engineer explainer on data labeling, the human work behind an AI model. It explains the general idea only and does not discuss the poll.

What the Poll Shows

Barber’s results card, titled “Private dataset buyers”, lists Surge with 4 votes, AfterQuery and Ooak Data with 3 each, and DataVendor, Grepped, Nyne and Standard Data with 2 each. The card says Grepped “received one negative vote as well”.

Eight more companies drew one vote each: Appen, Eigen Research, Fleet, Ionic, OpenAI’s data team, Patronus, Tenkara and Very Large Research. The card also names voters’ own companies that buy data, including BenchFlow, CodeMerchant, Convergent, Proximal, Snorkel and Zero Billion.

Barber writes that “People couldn’t pick a company they work at or founded”. His disclosure on the card reads: “on this list, I’m a small investor in AfterQuery, Convergent, Fleet, Ionic, Nyne, Ooak Data, Patronus, SF Data, and Snorkel. I advise Zero Billion”.

The post, above, had about 55,000 views when we read it on 30 October.

Who micro1 Is

micro1 said on 12 October that it had passed $1 billion in annual run rate and priced its Series B at a $4 billion valuation, as we reported. Chief executive Ali Ansari wrote: “today, i’m proud to share we’ve surpassed $1B annual run rate while remaining historically profitable”.

Three days earlier, micro1 announced a commitment to spend $1 billion over 12 months acquiring and licensing enterprise operational data, financed with capital provided by Citi and Hercules Capital, according to its release.

What Buyers Have Bid and Paid

The poll gives no prices, but our earlier reporting does. When Spirit Airlines’ data cache went to auction, Bloomberg reported bids of $7.5 million from Mercor, $10 million from Google and $12.5 million from micro1, as we reported in September. By our arithmetic, micro1’s bid was 67% above Mercor’s. The outcome of that sale is not established in the sources we used.

Grepped founder Erik Newsham wrote on X on 24 October that his company’s average deal is “nowadays” “closer to $1.1-$1.3m”, as we reported; he named no buyer or contract. Mercor chief executive Brendan Foody wrote that Mercor has “partnered with > 500 companies so far” for data.

DataVendor, a marketplace that opened a public beta on 22 October, prices data per environment, per task and per repo, and its homepage example cards put a payments monorepo with 2.4M lines of code at $220k, as we reported.

The Structural Read

As we read the results, the poll measures reputation rather than spending: it records which companies 36 people would point a seller to, not which buyer paid most.

As we read it, the list mixes data labs, marketplaces and a lab team: the card names OpenAI’s data team among the one-vote mentions, alongside marketplaces such as DataVendor.

By our arithmetic, the top two names, micro1 and Mercor, took 28 of the 64 votes listed, about 44%.

Three Implications

FOR SELLERS OF PRIVATE DATA The poll names 19 companies voters would call; our earlier reporting records micro1’s posted payment tiers and the Spirit Airlines auction bids.

FOR MICRO1 It drew the most votes, 19, and its own figures put it above $1 billion in annual run rate.

FOR BARBER’S PORTFOLIO COMPANIES SF Data, AfterQuery, Ooak Data and Nyne drew 5, 3, 3 and 2 votes; Barber discloses small stakes in each and says he did not vote.

The Business Engineer Lens

This story maps onto the Business Engineer framework The Four Intelligence Moats.

The framework puts it this way: “The Corpus Moat — built by pretraining, eroding as the public web is exhausted and synthetic data closes gaps.”

As we read it, the poll maps the buyers of data the public web does not hold: micro1 says its $1 billion commitment is for acquiring and licensing enterprise operational data. The framework was published in July 2026 and does not discuss micro1 or Barber’s poll.

What Is Not Established

The poll records which companies 36 people would point a seller to. It is not sales data or market share, and it gives no prices, deal counts or volumes. The card does not say whether a voter could name more than one company. Barber discloses investments in SF Data, AfterQuery, Ooak Data and Nyne, which drew 5, 3, 3 and 2 votes. The 30% share and the 67% bid gap are our arithmetic.

The Bottom Line

In Barber’s poll of 36 people in AI, micro1 drew 19 votes as the company to call to sell a private dataset, about 30% of the votes listed by our arithmetic. Its company figures put it at more than $1 billion in run rate, with a $1 billion commitment to buying company data.