The AI Talent Boom Is Measured by What People Put on LinkedIn

CBRE's count of AI-skilled workers has grown 131% in two years while the tech labor market containing them added roughly 3%, and the figure comes from profiles people fill in themselves.

New York passed the San Francisco Bay Area as North America's largest tech talent market this week, which is the sentence that ran everywhere from CNBC to Fortune. It comes from CBRE's Scoring Tech Talent 2026, the thirteenth edition of a report that has counted tech workers by metro for over a decade, and it is the first time New York has held the top spot. Nearly every version of the story attached the same explanation to it, which was AI.

The AI figures in that report deserve a closer reading than most of the coverage gave them.

Where the 751,000 comes from

CBRE puts the AI-skilled workforce across the US and Canada at 751,000 as of June 2026, up 45% in a year. That figure does not come from payroll records or from the Bureau of Labor Statistics. It is CBRE's analysis of LinkedIn Talent Insights, which aggregates profile data that members submit voluntarily, and CBRE's footnotes have said so for years. The methodology notes describe the AI counts as based on LinkedIn members who self-reported their occupation as tech talent with artificial intelligence and machine learning skills, and LinkedIn appends its own disclaimer noting that it cannot guarantee the accuracy of the data.

The AI talent population is therefore a count of people who have listed AI or machine learning skills on a profile. That is not a reason to throw the number out. Self-reported skills data is a reasonable proxy and often the only source available at metro granularity, particularly for a job category that no government statistical agency has a standard occupation code for. But it moves for reasons that have nothing to do with anyone being hired, and this week it was read as a hiring number.

The arithmetic does not work as hiring

CBRE put the US and Canada AI-skilled workforce at roughly 325,000 in mid-2024. A year later the figure was 517,000, which the firm reported as growth of more than 50%. This year it is 751,000. Across two years that is an increase of about 426,000 people, or 131%.

The tech labor market those workers belong to grew at a very different rate over the same stretch. US tech talent employment rose 1.1% in 2024, worth 64,140 jobs, and 1.8% in 2025, worth 108,760, which comes to something like 173,000 net new jobs on a base of around six million. Canada adds volume but not enough to change the shape of the comparison.

That leaves an AI-skilled population growing by more than twice the total net job creation of the entire market containing it. Both figures cannot be measuring hiring. The windows do not align perfectly, since the AI series runs mid-year to mid-year while the employment figures are calendar years, and anyone using the comparison owes readers that caveat. The gap is far too wide for the mismatch to account for it.

Reclassification is what closes it, and CBRE has described the mechanism openly for two years running. The 2025 edition attributed its 50% jump to employers redeploying and upskilling existing teams, calling the result a repositioning of talent rather than a hiring wave. This year's version says the Bay Area and New York each added more than 20,000 AI jobs since mid-2025 through a combination of new roles and existing jobs converted into AI-skilled ones. A data engineer whose title becomes AI engineer without changing desks adds one to the count.

Two different kinds of number got merged

The headline finding and the explanation attached to it rest on different evidence, and the coverage blended them into a single claim.

New York's 394,300 tech workers against the Bay Area's 375,730 comes from government employment statistics, and it is a hard count. New York gained 30,640 tech jobs between 2022 and 2025 while the Bay Area lost 23,900, so the crossover involves as much subtraction as addition, but the underlying data is solid. The AI explanation layered on top of it is the soft number.

The sectoral shift underneath both is the part that actually explains the change. Since 2022 the finance, insurance and real estate sector has added 90,530 tech jobs in the US while the high-tech industry has shed 21,262. Banks hire engineers, and banks are in New York. That accounts for the crossover without requiring a talent migration or the LinkedIn data at all.

There is also the ranking itself, which several outlets skipped. CBRE's weighted market score across thirteen metrics, the thing the report exists to produce, has the same top six as last year: San Francisco, Seattle, Toronto, New York, Austin and Washington. New York placed fourth on the ranking and first on one of its inputs.

Where this gets overstated

Upskilling is a real economic event. An engineer at a bank who learns to build and evaluate retrieval systems has genuinely changed what she can do, and a labor market where several hundred thousand people make that shift has genuinely changed. The problem is not that these workers are imaginary. It is that a conversion and a hire are being counted in the same unit and then reported with a verb that only fits one of them.

CBRE is not concealing any of this either. The methodology notes are published, the conversion language sits in the report body, and Colin Yasukochi, who runs the firm's Tech Insights Center, has been consistent in interviews that the Bay Area remains the industry's center. The qualifier survived the report and got lost somewhere in the summarizing.

The job postings data is firmer than the profile data and points the same way on geography. AI roles reached 31% of open US tech listings in June 2026, up from 11% at the mid-2022 peak, and 57% of listings in the Bay Area. Those come from Lightcast postings rather than member profiles, so they measure employer demand instead of self-description. They are still keyword-derived, which means an employer adding AI language to an otherwise unchanged job description moves them too.

On AI specifically the distance between the two cities is not close and is not narrowing. The Bay Area has 98,699 AI-skilled workers to New York's 67,949, and has taken 80% of US AI venture funding since 2020.

What the Bay Area number is actually saying

One figure in the report went almost unreported. Non-AI tech postings fell 60% nationally over the period and 73% in the Bay Area.

A region where 57% of tech listings are AI roles and the remainder has collapsed by nearly three quarters is converting into a single-sector economy at considerable speed. New York's total holds up partly because its tech workforce is spread across finance, media, health and advertising, which is a duller position and a more durable one. AI companies have also taken 58% of San Francisco office leasing in the first half of 2026, which is the same concentration showing up in the real estate.

Headcount comparisons will keep favoring the concentrated position for as long as the cycle runs. The more useful question the crossover raises is what happens to a labor market when 57% of what its employers advertise for is a single technology.