Close-up on a robotic eye. Digital illustration, 3D render.
getty
Over the past six months, job-cut announcements from big companies and a series of fear-inducing viral posts created a sense of doom about AI’s effect on the labor market. But on a recent trip to Rochester, to promote Capital Evolution, I had a chance to do my first real reporting about how AI is affecting entrepreneurship.
Journalists used to call this “gathering string,” finding the threads of the real story. In the entrepreneurial economy, AI is driving startup growth and some hiring, as a few successful companies start to emerge. It’s become much easier to bring launch products and companies. The AI boom is also evolving differently than the software boom. Software companies — and the men who almost always lead them — famously land millions or hundreds of millions of dollars in funding for untested ideas. In the AI economy, where ideas-in-digital-form are a dime a dozen, revenue and customers matter more. In other words, human skills are becoming more valuable.
The data that exists so far confirms what I heard from entrepreneurs on my panel discussion at the University of Rochester, and others in the local market. Startup rates remained at high level in 2024, according to the Economic Innovation Group. And economists, while noting that some job losses are showing up in early stage positions and in software jobs, say they have not seen a big labor market effect tied to the advent of AI, at least not yet.
Rochester Is A Deep Science Innovation Hub
The answer to the question of whether the next wave of deep science innovation emerges in America — which depends both on the science and an entrepreneurial base to bring the science to market — may emerge in Rochester and places like it.
That’s because AI may have enormous affects on a long-underfunded part of the innovation economy: Deep tech, or startups based on hard sciences. In those areas, AI makes data crunching easier and more accessible. Rochester is famous as the home of Kodak, Bausch & Lomb, and Xerox. The legacies of those giants made Rochester a center of optical technology.
“Every advanced semiconductor manufacturing system, whether it’s lithography, metrology, or advanced packaging, relies fundamentally on optics, lasers, and photonics. Without those technologies, you can’t make today’s chips—and in most cases, you couldn’t even make chips from decades ago,” said Joseph Spilman, CEO of Optimax Systems, which we wrote about in Capital Evolution. “For high-end semiconductor tools, there are roughly 10 organizations globally that can manufacture the required precision optics and optical systems—and two of them are here in Rochester: Corning and Optimax. For the most advanced systems, that number narrows further, and both companies remain in that top tier.”
AI’s impact is already visible beyond the lab. “In Rochester, it’s creating a whirlwind of economic development,” said Matt Hurlbutt, CEO of Greater Rochester Enterprise, pointing to a new data center and to the way economic development professionals are using AI to connect with companies considering the region.
At the University of Rochester, that shift is already visible. “Entrepreneurship has become a central theme across the campus,” said Roberto Colangelo, executive director of the Ain Center for Entrepreneurship and Innovation. “The appetite for entrepreneurship has never been stronger.”
That appetite is translating into activity that cuts across disciplines. “We have students working on projects that are landing on Mars… solving Alzheimer’s… growing tissues that will replace animal trials for the FDA,” Colangelo said. The scope reflects how AI is accelerating work well beyond software, particularly in deep tech fields where data analysis has historically been slow and expensive.
Down To Brass Tacks
On the panel, founders described AI in practical rather than abstract terms. Karla George, CEO of FLX AI, a consulting firm that is helping existing companies adopt AI, defined it simply: “It’s a technology that enables a machine, i.e. a computer, to mimic human intelligence.” That framing places AI on a continuum. “We’ve been consumers of AI for a very long time,” she said, pointing to recommendation engines and facial recognition as earlier waves.
What has changed is speed, and easier interfaces. George described a project analyzing geological risk data for pipelines that once took weeks. With AI, the same work can be completed “down to seconds and minutes.” The compression of time is reshaping how quickly companies can test and deploy ideas.
At NextCorps, an entrepreneurial support organization, the impact shows up in how companies get built. James Senall, president, said the barrier that once prevented nontechnical founders from launching has largely disappeared. “The question’s now, not if I can build it, but should I build it?”
He offered a recent example: “Thirty-five (teams) had fully functioning software… in a day and a half,” Senall said of a startup event where participants with little coding experience produced working applications. The bottleneck has shifted away from engineering. Perhaps, toward judgement about which problems are worth solving.
From an investor’s perspective, Rami Katz, CEO of Excell Partners, said the ability to build is no longer a differentiator. “It’s not enough to show that I can build it… yes, we know you can.”
The New Differentiator is Revenue
Instead, investors are focusing on traction and positioning. “How are you going to distinguish yourself from another thousand entrepreneurs that have the same idea?” Katz said. Access to customers and data, and the ability to move quickly with both, are emerging as the decisive advantages.
That reality is influencing where companies choose to build. Daphne Pariser, CEO of HeronAI, which won $1 million in funding from accelerator 43North for her company that automates financial reporting, argued that regional ecosystems can offer an edge. “Everybody has picked up my phone,” she said of building in upstate New York. “Either I’m happy to beta with you or I know somebody else who can.”
Those early customer relationships are increasingly necessary. “If you don’t have revenue, you won’t get funding,” Pariser said. The shift toward revenue and validation represents a break from the last startup cycle, when companies could scale valuations without clear business models.
The conversation also highlighted constraints that have not gone away. Data quality remains a central issue. George noted that many companies are eager to adopt AI without addressing underlying systems. “You have to have good data, you have to have quality systems in place,” she said, describing the frequent need to step back before deploying new tools.
Trust and deployment challenges are also front of mind. Pariser emphasized the importance of narrowing focus in companies that are adopting AI. “Make sure that you really know what the problem is you’re solving… and then ask them 15 more times,” she said. That discipline reduces risk when working with systems that can still produce unreliable outputs.
In regulated sectors, caution is built into the process. Katz described a “culture clash” between AI systems and industries like healthcare that require transparency. “AI brings tools and shortens decision-making pathways for the final human decision,” he said, emphasizing that full automation remains a distant goal in high-stakes environments.
The discussion repeatedly returned to human capabilities. As technical barriers fall, other skills are gaining importance. George put it directly: “The STEM and the tech, it’s table stakes.” Communication, critical thinking and domain expertise are becoming more valuable as differentiators.
What Are The Biggest Risks?
There are broader risks. Panelists raised concerns about data privacy, security, and uneven impacts across the workforce. Policymaking has not kept pace. “The legislators are still behind the Internet era,” Katz said, noting the difficulty of regulating technologies that evolve quickly.
Even so, the tone in the room was pragmatic rather than alarmist. Founders and investors are working through real use cases, testing products, and building companies. The conversation has moved toward execution.
The region’s industrial base, combined with a growing entrepreneurial network, is creating conditions for a startup ecosystem rooted in science and increasing speed.
The question of how AI will shape the labor market remains open. What is clearer, at least from this vantage point, is that the technology is expanding who can participate in building companies. The most important changes may not be about jobs lost or gained, but in the number of ideas coming to market. Optimistically, the quality of the ideas may be better, too.
