The San Francisco Bay Area, long the epicenter of the technology industry and now the heart of the artificial intelligence boom, is home to a vast concentration of workers employed by some of the world’s most valuable companies, including Apple and Nvidia.
But for the first time, New York has surpassed the Bay Area as the U.S. metro area with the largest tech talent workforce, according to a new report from commercial real estate and investment firm CBRE.
New York’s tech talent workforce reached 394,300 in 2025, compared with 375,730 in the Bay Area, according to CBRE’s annual Scoring Tech Talent report, which has tracked the industry for 13 years.
The shift reflects a broader transformation in the technology labor market. From 2022 to 2025, the Bay Area’s tech talent workforce declined 6%, while New York’s grew by more than 8%. The Bay Area’s decline comes as major technology companies have carried out large-scale layoffs while redirecting investment toward artificial intelligence.
CBRE defines tech talent broadly, encompassing highly skilled workers in more than 20 technology-related occupations across industries including technology, financial services, healthcare and government.
New York’s growth has been driven in part by the financial sector, which was an early adopter of artificial intelligence. AI startups have also expanded in Midtown South and other parts of the city. New York’s extensive public transportation network and large commercial and residential markets have further strengthened its appeal to both companies and workers.
“The sheer scale” of New York’s market supports early-stage growth while giving companies the ability to scale quickly, said Lauren Crowley Corrinet, vice chairman of CBRE’s Consulting Group in New York.
Still, the Bay Area remains the nation’s strongest overall tech market. CBRE’s ranking considers not only the size of the tech workforce but also factors such as salaries, housing costs and the supply of technology graduates. The Bay Area ranked No. 1 overall, while New York ranked fourth. Los Angeles ranked 18th.
Other U.S. markets are also benefiting from the redistribution of tech talent. Sacramento’s tech workforce grew by more than 8% between 2022 and 2025, reaching 42,970. The Los Angeles and Orange County workforce totaled 227,350 in 2025, up less than 1% from 2022. Toronto, Washington, D.C., and Dallas-Fort Worth also ranked among the markets with larger tech talent workforces.
The changes come as technology workers confront growing uncertainty about the impact of artificial intelligence on employment. A Pew Research Center survey released this week found that roughly 71% of U.S. adults expect AI to reduce the number of jobs in the country over the next two decades, up from 64% in 2024.
At the same time, the AI boom is creating demand for specialized workers, particularly data scientists and hardware engineers. CBRE executives expect those roles to continue growing as businesses across industries determine how to integrate AI into their operations.
“The Bay Area is likely to remain the central location for the AI industry and for innovation,” said Colin Yasukochi, executive director of CBRE’s Tech Insights Center. But, he added, as the industry expands, its growth tends to spread to other major technology markets.
For now, New York has claimed the title of America’s largest tech talent market by workforce size. But the Bay Area remains the country’s dominant technology ecosystem—and, according to CBRE, the leading market for tech talent overall.
Move Slow and Upgrade
Silicon Valley still worships “Move fast and break things.” Anthropic just admitted the world might benefit from slowing frontier AI. Then it kept racing. So did everyone else.
The mantra endures. Disrupt first. Monetize later. Externalize the wreckage. Critics pile on, yet the titans keep getting credit for “doing what American entrepreneurs do”—taking big risks, working hard, winning.
There’s a better path. We call it upgrading.
In *Move Slow and Upgrade: The Power of Incremental Innovation*, Albert Fox Cahn and I argue that real excellence often looks quieter. Upgrades are evidence-based. They set realistic expectations. They improve things one careful step at a time.
Engineers and scientists do this every day. So do teachers, tradespeople, nurses. Their patient work compounds into progress the flashier crowd rarely notices.
Look at the childproof medicine cap—simple, life-saving, unglamorous. Or Social Security, which became America’s largest government program through deliberate, incremental expansion, not overnight revolution.
Silicon Valley treats upgrades like allergies. It leaps. Consumers pay.
Tesla led the charge to kill analog knobs for touchscreens on stereos and climate controls. Sleek. Dangerous. Drivers looked away from the road. Other makers copied. Common sense finally fought back. Knobs are returning. Consumer Affairs called it one of the biggest reversals in recent automotive design.
Real upgraders improved cars differently: backup cameras, blind-spot alerts. Safety gains without yanking the driver out of control.
Zoom out. The metaverse. Zuckerberg sold it as the next great human connection, then renamed his company around the bet. Result: at least $80 billion vaporized. No clear definition. Expensive, nausea-inducing hardware. No killer app. An upgrader would have spotted the red flags on day one.
Amazon’s Ring cameras and Neighbors app promise high-tech safety through remote monitoring and community alerts. Upgraders ask a simpler question: Do locks, lights, and sensors deliver better returns with fewer privacy costs? Evidence that doorbell cameras actually cut crime remains thin. The tech often sells the illusion of security.
Even Silicon Valley’s “sensible” moves can overshoot. AI scribes listen to doctor visits, transcribe, and draft notes. Paperwork is crushing physicians. Automating it looks like a perfect small upgrade—far better than the fantasy of replacing doctors with bots. Yet adoption is racing ahead of evidence.
Dr. Benn Gooch tried them, then stopped. The tools worked as advertised, but something eroded: clinical memory, narrative sense of the patient, the cognitive work of thoughtful documentation. Researchers warn that risks remain understudied, governance thin, and safer designs (ones that keep doctors more involved) untested.
The same pattern hits policing. Tools like Axon Draft One promise faster, more consistent reports. Independent research raises flags on accuracy and efficiency. There’s also the subtle risk that officers start remembering what the AI wrote, not what they saw.
None of this bans the technology forever. It demands we earn the upgrade—with data, with time, with caution.
Haste is not progress. Measured change is. Upgrading is how we actually get better without breaking what still works.