
What happened
Startups including Handshake, Micro1, and Mercor are pitching companies on selling internal work files—documents, emails, Slack messages, tickets, and call transcripts—for $350K to $1.8M, with one pitch reaching $2 million, to train AI models.
Why it matters
This data-acquisition business is way bigger than it appeared, and its practitioners say it could have as much impact on workers and society as the data centers dominating the news.
Summaries like this, every morning.
Mercor originally started as an AI recruiting company, but as the team got to know the players, it learned that the big labs were frustrated by the quality of data. The industry had been hiring experts to simulate actual workflow, but that didn't capture the messiness of an actual operating business. That gap helps explain why buyers are now turning to live corporate records rather than synthetic or simulated work.
Handshake tells WIRED that a big part of its business is buying graduate students' papers and coursework, while an entrepreneur described mail from a company named Sunset that helps startups wind down and brokers the data to labs. Micro1 CEO Ali Ansari says labs, enterprises, and robotics companies pay about $40 to $50 billion a year for data, a figure that could not be verified. He expects that to exceed a trillion dollars a year in the next two to four years.
Anthropic acknowledges a big market for data but told WIRED it doesn't buy workflow from live companies like WIRED; OpenAI didn't respond before publication. The Spirit Airlines bankruptcy sale, where a Google bid of $10 million was reportedly topped by Micro1 by $2.5 million before the sale was put on hold, shows how contested these deals can become when worker privacy is at stake.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Vals tested AI models on simulated corporate networks; OpenAI's restricted Daybreak completed 81% of tasks, wh…

Meta introduced Spaces, a sidebar feature in its Muse agent interface that builds a tracker, planner or projec…

Neil P. Osnato, founder of Persistence Analytics Group, published an argument that AI infrastructure commitmen…
Atomicwork launched an AI Coworker Marketplace with certified AI coworkers for cloud infrastructure, access ma…
Tech stocks rose the day after the midterms, then plunged Thursday for a collective $2.5 trillion market-cap l…
Intercontinental Exchange CEO Jeff Sprecher told Semafor he is "bullish but cautious" on trading GPU compute…
