Bill Gross founded Idealab in 1996 as a startup incubator, and he has long treated coming up with business ideas as his particular talent. In a video played on the Moonshots podcast, Gross said AI had recently done something that "stopped me cold": it could outperform his own "superpower" at exactly that. He decided to scale the ability up and give the results away. His new site, Idealab.ai, offers more than 1.2 million AI-generated business ideas, and anyone can search them for free.
Host Peter Diamandis, founder of XPRIZE, brought up the launch on the episode, which was recorded on October 20, 2026 and published the next day. He said Gross had launched the site "yesterday" and called him "a close friend and a friend of the pod." When Diamandis first introduced the platform he said it held 1.2 billion ideas. That was a slip: he used the 1.2 million figure later in the conversation, and the site says more than 1.2 million. Gross's own video said "more than 1 million."
The launch set off a debate that ran through the rest of the discussion: if ideas cost nothing, what becomes scarce?
How the idea engine works
Diamandis said the system analyzes more than 17 million sources and applies 31 methods Gross developed for spotting opportunities, including unmet needs, new technologies, regulatory shifts and cultural changes. On the site's explanation page, Gross says he documented those 31 methods and refined them with AI. Because the system turns up far more opportunities than Idealab could ever pursue, the public ideas are free to explore and use, with no requirement to buy the company's services.
According to Idealab.ai, the engine reads research, patents, news and customer complaints. It sorts what it finds into three kinds of signal: an unmet opportunity, a new capability that makes something possible, or a structural shift. Then it combines signals into business concepts. The site's example pairs one complaint, that people struggle to work out which skincare products irritate their skin, with a newly cheap capability, automated reading of ingredient labels. Each resulting brief lists supporting sources, startup-cost and revenue estimates, a roadmap for the first 90 days, competitors and risks. A chat tool lets users tailor the brief.
Users can enter their own skills, interests, available money and location, or type in an idea they already have to find related or competing concepts. The site's filters cover startup capital, time to market, team size, delivery type and market reach. One preview brief proposes using new maps of the nose's odor receptors to screen aroma chemicals by computer.
Diamandis said Gross's goal is a million new businesses and 100 million new jobs. He also read out a line he attributed to Gross in Fortune: with Idealab "I could do 10 companies at once. Now I can do 10,000 at once." Diamandis said Cloudflare had committed $100,000 in credits to each startup built on the platform, and Idealab.ai's business-tools page confirms that Cloudflare is offering $100,000 in credits to businesses built there.
For Diamandis, this is "the abundance thesis at large." Business ideas are no longer scarce, he said, so "scarcity moves to execution, taste, and courage." He backed this up with startup costs, citing the investor Mark Suster of Upfront Ventures as saying a tech startup cost about $5 million to launch in 2000 and about $50,000 a decade ago. Today, Diamandis added, it takes "a couple of K." Suster's 2011 article supports the falling trend. Suster recalls spending $2.5 million on infrastructure and another $2.5 million on staff and commercialization for a company he started in 1999, and explains how open-source software, cheap interchangeable servers and Amazon's pay-as-you-go cloud computing brought those costs down. The "couple of K" figure is Diamandis's own estimate for today; the article, written 15 years earlier, does not contain it.
"A dime a dozen"
Diamandis asked Chamath Palihapitiya, founder and CEO of Social Capital and co-host of the All-In podcast, whether a free list of ideas helps founders or whether the idea is the least valuable part of a startup.
Palihapitiya was unimpressed. "I don't think ideas are all that valuable. I think they're mostly a dime a dozen," he said. "I don't think I've had an original idea my entire life." Someone has always had it before, he said. What matters is the difficulty of starting, of putting one foot in front of the other, and of working through the low points. That "has nothing to do with an idea. It has everything to do with your makeup." Palihapitiya said he has been through many failures. A good idea does not show whether a person can manage through them, and that, he said, is "the key."
A tick list for people who are stuck
Salim Ismail, founder of Open ExO and a general partner at Exponential Venture Capital, said he was "very, very excited." Investors like Palihapitiya are "swimming in ideas," Ismail said, but many people are stuck, unsure what they want to do. For them the site is a tick list: does this excite me, or this?
Ismail went through the list and picked four favorites, each aimed at the scale of a planet or a whole society. A "global abundance grid" would treat physical resources as fluid assets and reroute them algorithmically. "Post-bureaucratic liquid government protocols" would replace representative parliaments, which Ismail argued were designed for a world where information was scarce, with real-time polling of citizens. The other two were an open-source planetary research and development network and algorithmic conflict resolution, an open-access layer of judicial infrastructure built into the internet in place of pay-to-play legal processes.
Ismail also pictured the list as input for AI. He suggested handing it to AI systems, including what he called "our Argentinian AI native organizations," with instructions to try the ideas and see which work, at very low cost. He called this taking "lean startup to the next level." Lean startup is a popular method of testing business ideas through quick, cheap experiments. With 1.2 million ideas, Ismail said, that means 1.2 million experiments that could be run and modeled.
His enthusiasm had a personal source. Ismail said he did not grow up in an entrepreneurial family: his mother was a German professor and his father an engineer. Only when he took a course on technical entrepreneurship at the University of Waterloo did he realize, "oh, you can actually build a business? Like, that's a thing?" It took him another 20 or 30 years to catch "the full entrepreneurial bug." With a list like this when he was younger, he said, he would have picked one idea and run with it. He thinks it gives a starting point to young people who are lost.
Diamandis tied the site to a recurring theme of the show. In his view, the old social contract of good grades, a good college and a degree "is cooked or going to be cooked," and it matters more to find something purposeful and build. For people who don't know what they want, he said, the site is at least a "shopping mall" of ideas. He added that people can do much of this on their own: take a problem you care about, or "something you hate that exists," give it to your favorite large language model, and ask it to help you come up with an idea, design a business and build a prototype.
A "baby step" toward businesses run by AI
Diamandis then asked Alexander Wissner-Gross, a computer scientist and founder of Reified, how he had been involved. Wissner-Gross began with a disclosure: "I have a financial interest in the venture behind this."
More than a year ago, he said, he told Gross that within a year or two AI would be able not only to come up with good businesses but also to execute on many of them. "He didn't believe me at first," Wissner-Gross said. A year later Gross started to believe, because the capabilities kept improving. Idealab.ai is the result. Wissner-Gross said he was limited in what he could say but that there was "much more to come."
He placed the site within a bigger argument. Reaching what he called "universal high income" will require, to some extent, AI that makes ideas "too cheap to meter" and makes carrying out those ideas too cheap to meter as well. He expects "a proliferation" of companies or business units that are AI engines from start to finish, handling ideation, management and execution. If machines end up doing most of the thinking, he said, it would make little sense for humans to run most businesses, though he added that he did not expect that scenario to come true. He called Idealab.ai "a baby step" toward businesses that machines both conceive and operate.
The company's own plans point the same way, with people still involved. Idealab.ai's catalog divides running a business into 118 elements, from customer research and accounting to procurement, fundraising and executive judgment. It plans to connect existing services where they exist and build the missing ones. The first five elements are due to launch first, with more added during 2027. AI agents, meaning software that carries out multistep tasks using tools, would handle the operational mechanics, while people set goals and limits and give approvals. Gross's explanation page says the company is building tools for capital access and agentic execution, while learning from customers, adapting and judgment stay with the entrepreneur.
Why give it away
Dave Blundin, founder and general partner of Link Ventures, said he sits on the board of Prorata AI and sees Gross often. He thought the free giveaway made commercial sense. Some ideas on the list can be pursued with almost no budget, he said, but many more need funding and computing power. People will find a good idea, then come back to Gross for help getting funded or getting compute: "Hey, I want to execute on idea number 800,462 help." Blundin called it "a really good business plan" and said he now understood "why he open sourced the whole thing."
Diamandis agreed, describing the site as "a great magnet for entrepreneurs," especially if the goal is to "collect all their emails and their contact details."