$500 million a year is not a side hustle. It is a pipeline.
Top US data companies supplying data to Chinese AI — while also working for Pentagon, GOP rep warns
As an Amazon Associate I earn from qualifying purchases.
$500 million a year to Beijing's labs. The same shops still invoice the Pentagon.
Summary
- The New York Post reports Rep. Michael McCaul warned that four U.S. AI data firms, SurgeAI, Mercor, AfterQuery, and Turing, help label and package training data for American labs and the Pentagon while also serving Chinese AI labs.
- Forbes, cited by the Post, says China's top six AI labs spend about $500 million a year on American data companies.
- SurgeAI has done work with Tencent, which the United States has designated a Chinese military company, and also holds small Army and Air Force contracts.
- McCaul said firms that want U.S. government work should not help a Pentagon-designated Chinese military company build better AI.
Commentary
A night-shift electrician still tapes a conduit in a server hall his taxes already cooled. He did not buy that power so a vendor could invoice Beijing before dawn.
That is the gap. Four shops. $500 million. Tencent on the designation list. The honest dad already paid the contract. The lab paid twice.
Look every sailor who still trusts a model in the eye and answer this: if the same labeler feeds the Pentagon and a PLA-tied lab, who still owns the dataset?
Comments
I signed a clearance. A dual invoice is a breach of common sense.
Tencent is designated. SurgeAI has Army and Air Force work. Pick one customer.
Allied vendors already face a China clause. Copy the discipline.
SurgeAI, Mercor, AfterQuery, Turing. Four names. One rule.
I still cool a hall with overtime. I do not cool a Beijing model.
A kid's future job should not train the other side's model.
A republic that sells its labels will sell the next war.
Post printed $500 million, four firms, Tencent, and McCaul. Argue those nouns.
Cut the China invoice. Keep the contract. A dataset is not a souvenir.