On June 25, 2026, the four companies racing hardest to build artificial intelligence signed their names to the same uncomfortable admission: the technology they are building is going to push millions of Americans out of work, and someone has to pay to move them somewhere else. The vehicle is RAISE US, a bipartisan nonprofit co-chaired by former Commerce Secretary Gina Raimondo and former Indiana Governor Eric Holcomb, anchored by the OpenAI Foundation, Anthropic, Microsoft, and Amazon. It launched with more than $500 million in hand against a $1 billion target. The people funding the cure are the people who built the thing it treats.
The industry finally said the part it had been avoiding
For three years the standard line from the largest AI labs was that their models would augment workers, not replace them. RAISE US is the quiet retirement of that line. The nonprofit estimates that 50 million American jobs are exposed to AI, and it exists because its founders have stopped treating that exposure as a thing that might happen.
Listen to how the backers talk now. Microsoft vice chair and president Brad Smith, describing how a single AI agent can swallow a white-collar task, did not hedge. "You can think of doing that with almost any job we have," he told the Philadelphia Inquirer. That is not a warning from a critic. It is a description from one of the people building the tools.
Raimondo was blunter. "If we build the best AI systems in the world and leave millions of Americans behind, we won't have won anything; we'll have automated our own decline," she said at the launch.
The effort is real, and that is what makes it hard to dismiss
None of this is a press release dressed as action. Four states are already running pilot programs with the backing of their own governors: Utah, Arkansas, Maryland, and Connecticut, testing wage support, retraining incentives, and AI career coaching. The political coalition is genuinely mixed, pairing Republicans Sarah Huckabee Sanders and Spencer Cox with Democrats Ned Lamont and Wes Moore.
Its backing runs wider than the four anchors too, taking in Bank of America, General Motors, IBM, and the Rockefeller Foundation. And labor has a seat: AFL-CIO president Liz Shuler sits on the advisory board, and has framed her role as bringing the voice of the labor movement into a conversation that usually happens without it. For an industry that tends to treat unions as an obstacle, putting one at the table is not nothing.
Hold that thought, because a seat and an agenda are two different things.
A response designed by the people who profit from the problem
Here is the structural fact the launch cannot dissolve. The companies funding the response are the same companies shipping the products that make the response necessary, and money shapes which solutions get tested and which questions never come up.
Look at what the pilots actually fund: wage support, retraining incentives, career coaching. Every one of them treats displacement as a problem the worker has to solve, with help. None of them touches the questions a differently-funded effort might reach for first. Should deployment move slower in the sectors least able to absorb the shock? Could the firms capturing the productivity gains pay a direct share to the people losing the jobs? Those options are not on the table, and they were never going to be, because the table is set by the balance sheets that benefit from leaving them off.
Labor has a seat. It does not have the pen.
A $1 billion answer to a $725 billion question
Then there is the matter of scale. RAISE US is targeting $1 billion in multi-year commitments and has secured a little more than half so far. Set that beside what the same tier of companies is spending on the other side of the ledger. Microsoft, Amazon, Alphabet, and Meta are on course to spend roughly $725 billion on infrastructure in 2026 alone, the bulk of it for AI, up about 77% from the year before. Two of those four, Microsoft and Amazon, are RAISE US anchors.
The arithmetic is stark. For every dollar this cohort has pledged to worker adaptation over several years, it plans to spend on the order of seven hundred building the systems in a single year. One side of that equation compounds at the speed of software. The other moves at the speed of a human being learning a new trade. Good faith does not change which one is faster.
Why it matters
Strip away the specifics and a template is taking shape. When a disruption grows large enough to frighten its own creators, and the creators move first to fund the response, they also get to define what the response is. RAISE US defines it as a worker-adaptation problem: the jobs are changing, so change the workers. That framing is not false. It is incomplete in precisely the ways that happen to be most expensive for the companies paying for it.
The sincerity is real, the leaders are credible, the pilots are running, and the structure still routes every hard question around the people who profit most from never asking it. A bipartisan nonprofit with a former Commerce Secretary at the helm and a union president on the board is the most legitimate possible vehicle for that outcome. Legitimacy is what makes it effective. It is also what makes it worth watching closely.
The wave and the lifeboat are being built by the same hands. So who should decide how big the lifeboat needs to be, the people losing their jobs, or the people whose business depends on building the wave faster?
Originally published as an Instagram carousel on @recul.ai.