The bank's new Small Business AI Institute pairs Wall Street money with Anthropic, Google, Meta, and OpenAI, even as David Solomon plays down the need for frontier models
Goldman Sachs is putting $500 million behind a new effort to help US small businesses actually use artificial intelligence. It brought four of the biggest names in AI along for the launch.
Then CEO David Solomon told the audience that most of the economy doesn’t need the most powerful models those companies build.
The bank unveiled its Small Business AI Institute on Thursday, October 22, 2026. The event took place at Goldman’s headquarters in lower Manhattan, and Fortune was in attendance.
The institute carries a $500 million commitment. It is designed to strengthen 10,000 Small Businesses, the bank’s long-running entrepreneur program.
“But there’s very little that we do where we need frontier models,” Solomon said.
A Babson College survey of program alumni, conducted in August 2026, found that 86% now use AI in their operations. Two years earlier, in 2024, that figure was 56%.
Only 15% of respondents said they had fully built AI into their core business functions.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Goldman launched 10,000 Small Businesses in 2010. Since then, it has served nearly 20,000 entrepreneurs across every US state and territory.
The bank has committed a total of $1.25 billion to support small businesses. Graduates have added jobs four times faster than comparable US businesses. Together, alumni companies generate more than $32 billion in annual revenue and over 350,000 jobs.
The Babson survey itself reflects that advantage. Goldman can track AI adoption within a defined group over time, which most AI training efforts cannot.
For the AI labs, the partnership is about distribution. Anthropic, Google, Meta, and OpenAI spend heavily building ever-larger models, but revenue depends on businesses actually using them.
Solomon’s comments add an interesting wrinkle to that pitch. If most small-business tasks don’t require frontier capability, then competition shifts to cost, ease of use, and reliability.
The capital component deserves close attention. Training helps owners understand AI, but integration often requires spending on software, data cleanup, and staff time that small firms struggle to fund.
The key metric to watch is the 15% figure. If future alumni surveys show full integration climbing meaningfully, the institute will have a strong case that it changed how small companies operate.