Bill Gross wants Fortune 500 companies to pay $250,000 a year to mine their patents with AI
Idealab founder Bill Gross wants large companies to pay $250,000 a year to run his AI system on their own patents. He made the pitch at the Fortune 500 Innovation Forum in Detroit on November 16, 2026. A month earlier, he gave away 1.2 million AI-generated business ideas for free. Now he wants companies to pay for a private version built on their patents and trade secrets.
Key points from Gross's pitch
- Idealab.ai plans to license its full system to companies for $250,000 a year.
- Five companies can join a one-year pilot for $25,000.
- Gross says companies would run the system on their own servers, so private data stays in-house.
- The $250,000 licenses are meant to fund the free public version for entrepreneurs.
What Gross told the executives
Gross asked the audience to think of their own companies as the input for the AI system. “What if the input isn’t the news? What if the input is you?” he said, referring to their patents, trade secrets, and shelved research. Idealab.ai ran the public patents of 113 companies attending the forum through the system and handed each one a printed sheet of business ideas they were not already pursuing. Gross called that a teaser and said he wants companies to share their non-public information too.
How the system was built and tested
Gross launched Idealab.ai in October. He fed it news stories going back to 2000, every patent since 1980, product recalls, complaints on Reddit and X, and his own 31 methods for generating business ideas. He said it produced 1.2 million ideas after about a month and roughly $1 million in AI costs. He also said 131,000 people in 202 countries have searched the ideas since launch.
To test a corporate version, Gross turned to Caltech, where he is an alumnus and trustee. He asked the tech-transfer office for four patents. He said the office replied, “Why just four? We’ll give you all 9,000.” He said the system processed all 9,978 patents and generated 3,867 businesses that Caltech should either build or license. For comparison, Caltech's tech-transfer office signed 59 licenses and options and helped form eight startups in fiscal 2024, according to its annual report.
Researchers question the premise
The pitch targets a real problem. Large companies hold many patents that never become products. Research by Kellogg's Sara Moreira and economists at Yale and the Atlanta Fed, which matched about 1.4 million products to patents, found that large firms' patents are tied to new product launches far less often than smaller firms' patents. But Moreira's explanation works against Gross's pitch. She found big companies often use patents defensively, to block competitors, rather than to launch products. Some shelved research is shelved on purpose.
What is confirmed and what is not
Confirmed details include the pricing, the pilot offer, the Caltech test, and the forum demonstration, all reported from Gross's on-stage remarks and the Fortune report. The idea counts, the $1 million in AI costs, and the user figures come from Gross himself and have not been independently verified. It is also not confirmed that any company has signed up for the license or the pilot.
Why it matters
Big companies spend heavily on research, and much of it never reaches the market. If Gross's system works as he describes, it could give firms a way to find new uses for patents they already own. The test is whether corporate research departments, which have had decades to find uses for their own patents, will pay him $250,000 a year to do it for them.