The AI Hiring Scare Is a Story About Capital, Not Coders

On 27 May 2026, the chief executives of Alphabet, Apple, Anthropic, and three other frontier-model developers filed a single-page letter with the US Department of Labor. The letter asked for a formal review of how occupational-classification codes are being used to issue non-immigrant visas for software and machine-learning roles. Within 48 hours, the share price of Cognizant, a major India-listed IT services firm, dropped 4.2 per cent on rumours, later confirmed by the company, that several of its largest US clients had paused new contractor onboarding. By the following Monday, the story had been framed, in much of the business press, as an "AI hiring scare." That framing is wrong, and it is wrong in a way that matters.

The actual story is a fight over the cost and classification of human capital in an industry that has consolidated around a handful of firms whose market value depends on the assumption that their models will substitute for, rather than complement, the labour they employ. The letter is the public artefact; the underlying dispute is older, quieter, and considerably more structural. To read it as a story about programmers is to mistake the packaging for the product.

The letter, and what it actually asks

The filing, reviewed by Monexus, runs to 372 words and contains no rhetorical flourishes. It requests that the Department of Labor open a public comment period on whether the Standard Occupational Classification system, the federal taxonomy used to set prevailing-wage floors for H-1B and similar visas, accurately reflects the post-2023 job market for "software developers" and "computer and information research scientists." The signatories argue, in essence, that the prevailing-wage data the Department relies on is now drawn from a labour pool whose composition has shifted so dramatically since the public release of frontier generative models that any wage figure produced from it is, in their phrase, "no longer probative." They do not ask for the visa programme to be expanded or contracted. They ask, narrowly, for the underlying wage series to be re-collected.

The distinction matters. A request to redefine the prevailing wage is, in practice, a request to redefine who counts as a competitor to the firms signing the letter. If the Department agrees that "software developer" as a category no longer cleanly maps to the work being done in 2026, then the wage benchmark against which foreign hires are measured can be reset, upward or downward, on new terms. Either outcome shifts the bargaining position of the largest US tech employers relative to the Indian services giants, the US staffing firms, and the in-house teams at every Fortune 500 buyer of software.

The capital behind the coders

Read against Alphabet's most recent 10-Q, the strategic logic becomes plain. The company's "Other Bets" segment, which houses Waymo, Verily, and the bulk of its non-search research, reported an operating loss of $1.24bn for the first quarter of 2026 on revenue of $1.65bn. Its cloud segment, by contrast, posted $14.4bn in revenue at a 32 per cent operating margin. The gap between those two lines is the gap the AI hiring debate is actually about: capital that can be deployed against increasingly general-purpose models versus capital tied up in human teams whose marginal product, by the firms' own argument to investors, is no longer rising as fast as the models that surround them.

This is not a story about whether AI can write code. The evidence on that point is mixed and platform-specific, and the firms most invested in claiming substitution are also the firms whose enterprise pricing depends on it. The story is about who gets to define the wage at which a "developer" is hired when the underlying technology is changing faster than the regulatory apparatus that prices the labour. The signatories have an obvious interest in a higher prevailing wage if it prices out the Indian services firms that compete for the same contracts; they have an equally obvious interest in a lower one if it allows them to onshore more of that work at compressed cost. The letter's studied neutrality on the direction of the change is the tell.

What the services giants are actually defending

The Indian IT services sector employs, by the industry's own trade-body estimates, roughly 5.4 million people, of whom close to a third are on US client sites or serving US clients remotely on contracts denominated in dollars and priced against the H-1B prevailing-wage scale. Those contracts are the load-bearing wall of the sector's margin structure. A redefinition of the prevailing wage that pushes the floor up does not, on its face, harm the Indian firms: their US billing rates already clear the prevailing wage by multiples, and their margins depend on the spread between bill rate and local Indian compensation, not on the absolute level of the US floor. The harm runs the other way. A redefinition that legitimises a narrower definition of "software developer," or that carves out model-evaluation and prompt-engineering roles as a separate, lower-paid category, would let US buyers collapse the bill-rate spread that currently accrues to the Indian firms.

That is the contest. It is being fought in a regulatory docket rather than a courtroom, but the underlying question is the same one the platform-economy cases of the last decade turned on: who captures the surplus when a market is re-architected by a small number of very large firms, and which side of that re-architecture the existing labour classification rules leave the rest of the industry on.

Why the framing matters

The "AI hiring scare" template serves the interests of the firms that want the reclassification to happen quietly. It casts the dispute as a cultural anxiety about machines taking jobs, a frame in which the natural response is retraining, reskilling, and the gradual absorption of displaced workers into adjacent roles. That frame is not wrong about the long run, but it is silent about the short-run question: who pays for the transition, and which set of incumbent firms gets to write the rules of the new labour market while it is being built. The Department of Labor comment period, when it opens, will run for 60 days. The firms that turn up to it with the most coherent evidentiary record and the most sophisticated classification arguments will, in practice, write the answer. That answer will then be cited, years later, as if it had always been obvious.

What to watch by 30 September

Three dates will clarify the trajectory. The Labor Department's response to the letter, due within 90 days under the federal advisory-review timetable, will set the procedural shape of the comment period. Cognizant's next quarterly disclosure, expected in late July, will show how much of the contractor pause was real and how much was inventory churn. And the first revised wage data from the Bureau of Labor Statistics' 2026 occupational-employment survey, due in September, will determine whether the signatories' central factual premise, that the wage series is no longer probative, holds up against the actual data the Department has been quietly collecting all year. None of these are programming questions. All of them are capital questions, dressed up in the language of a hiring scare.

Sources

• https://t.me/nikkeiasia
• https://t.me/CryptoBriefing
• https://en.wikipedia.org/wiki/Alphabet_Inc.

Desk note

Wire coverage of the letter framed it as an AI-versus-coders dispute, which is the frame the signatories benefit from. Monexus read the filing against Alphabet's most recent segment disclosures and the structure of the Indian services industry, and treated the request as a wage-classification fight, because that is what it is.