A new working paper from Harvard Business School and INSEAD finds AI-native startups are building smaller, flatter teams with fewer entry-level workers. The study of Y Combinator startups from 2020 to 2024 also finds a higher share of senior talent at these fi
For years, the AI boom has carried a promise: the entry-level ladder would become less of a bottleneck. New tools could let more people move faster, learn on the job, and take on bigger responsibilities sooner.
But a working paper from Harvard Business School and INSEAD points in a different direction inside AI-native startups—firms that build their product and their internal work around AI.
The researchers examined startups affiliated with Y Combinator from 2020 to 2024. as well as a broader set of US venture-backed startups whose first financing closed during the same period. Their core conclusion is blunt: AI-native startups are building smaller. flatter teams and hiring fewer entry-level workers than their non-AI peers.
The paper, titled “AI-Native Firms,” defines an AI-native startup by two shifts in productivity.
First is the “process channel.” These companies use AI inside the company to make employees more productive—helping them code, sell, design, or coordinate work faster.
Second is the “product channel.” They also embed AI directly into what they sell, enabling customers to use the product to perform work that once required human teams.
Those structural choices show up in the staffing numbers. The study finds AI-native startups are 25% smaller. They have about 13% more engineers, and their shares of entry-level workers and managers are each roughly 15% lower than non-AI-native startups.
That challenges a widely accepted premise of the AI boom—one where AI reshapes the bottom rungs of the career ladder. The authors acknowledge how that idea took hold: entry-level workers can use AI to take on bigger responsibilities sooner and automate routine tasks. At the same time. “vibecoding” has made it easier for non-engineers to turn ideas into prototypes. blurring the threshold and need for technical talent.
Yet the paper’s findings suggest the demand created by these changes doesn’t flow evenly downward. Instead, AI-native startups appear to draw more talent from the expert end.
The share of senior workers at AI-native startups is 20% higher than among non-AI-native startups. The researchers also report that these firms attract a specific type of worker.
“AI-tagged firms employ smaller teams of more talented and technical workers. These workers are especially likely to be graduates from elite institutions, concentrated in Silicon Valley, and male,” the authors write.
The numbers therefore land with a particular weight: if AI tools do accelerate learning, the paper suggests they may not be democratizing opportunity so much as concentrating it.
The authors’ bigger concern centers on what happens over time when adoption—and the benefits that come with it—don’t spread evenly. Their warning is explicit.
“If AI tools accelerate learning for those who use them, differential adoption rates may translate into widening performance gaps — both for individual workers within firms and for the entrepreneurs who found them,” the researchers write.
Taken together. the pattern the study documents is hard to miss: smaller teams. fewer entry-level hires. and a workforce that skews toward already-credentialed. highly technical talent. In the race to build with AI. the career ladder inside these companies may be moving—but not toward the middle of the market.
Whether that becomes a temporary phase of hiring as companies grow, or a lasting shift in how opportunity is allocated, is now the question sitting behind the data.
AI-native startups Harvard Business School INSEAD Y Combinator entry-level hiring venture-backed startups workplace productivity elite graduates senior talent labor market