Mark Cuban on Why AI Could Turn Solo Founders Into Trillionaires

Mark Cuban’s central prediction is that a solo founder directing a swarm of AI agents could become a trillionaire. He sees AI as a technology that can let one person do the work of a much larger organization, but not as a simple replacement for human judgment: people who know how to use AI may replace those who do not. Models still need context, oversight and decisions about which questions matter. Cuban argues that companies should teach employees and executives how agents work, including latency and the differences between open- and closed-source models. He calls AI the greatest transitional technology he has seen and expects businesses no one has yet imagined to emerge as people become more confident using it.

For entrepreneurs, Cuban’s advice is to use AI to identify unsolved problems and explore possible opportunities, while treating its answers as something to check rather than accept. He compared Grok and ChatGPT while reviewing contracts and stressed that people should use AI to learn, not to avoid learning. He sees becoming comfortable asking questions by voice or text as a key hurdle. He also argues that AI companies need to understand social-media algorithms and reach people through the feeds they actually see; people encounter different versions of the world, making broad position papers and conventional advertising poor tools for changing minds. Opposition to data centers, he added, can reflect either misunderstanding of AI or dissatisfaction with the deals made around them.

One near-term healthcare use Cuban sees is reviewing insurance and pharmacy-benefit contracts before employers sign or renew them. AI can flag terms in agreements involving benefits, brokers, consultants, third-party administrators and pharmacy-benefit managers, giving employers information to negotiate. He pointed to the City of Dallas’s 1,335-page pharmacy-benefit contract as an example of what may be hard to scrutinize manually, while warning that AI cannot undo unfavorable terms after an agreement is signed. Julie estimated that about 155 million Americans, roughly 60% of the country, have employer-sponsored coverage and said some carriers had imposed year-over-year premium increases of up to 60%, including for startups. Cuban said about 60% of people receiving company healthcare are covered by self-insured employers, and characterized medical claims administration as the larger problem: roughly 90% of the market, compared with 10% for pharmacy benefits.

Cuban cautions that AI will not fix healthcare’s underlying economics or create price discovery by itself. He describes a market shaped by regulatory and scale capture, where large vertically integrated companies can use technology to improve earnings and shareholder returns without lowering costs for employers and patients. He said UnitedHealth has more than 3,000 subsidiaries and $150 billion in annual inter-company transfers, and argued that self-insured employers may lack the claims data needed to reconcile what they pay with what providers receive. His proposed transparency measures include making taxpayer-funded contracts public without redactions, allowing patients to count a lower price they find toward their insurance deductible, and requiring nonprofit hospitals to publish detailed general ledgers and procurement receipts. He said eight states already let patients apply a lower price toward a deductible at some level.

Cost Plus Drugs, which launched on January 19, 2022, with 111 generic medications, has grown to roughly 4,000 SKUs, according to Cuban. It uses a stated 15% markup, avoids marketing and advertising, and has no agreement with the big three pharmacy-benefit managers; he said at least one company had been told it was part of a big-three network. Cuban said publicly available prices have prompted incumbent responses, while the company’s biggest challenge is keeping up with week-by-week growth. AI and robotics also let the company fit manufacturing into a 5,000-square-foot Dallas facility, where conventional production would have required a football field or more. Its line can produce a six-month supply, turn over every four hours and switch drugs.

Cuban said the company’s manufacturing pods cost about $20 million each; the government received its first pod and ordered a second, and he expects additional orders. He sees pods as a way for research hospitals to make and quickly iterate N-of-1 therapies on site. Scaling domestic production remains constrained by costs: each generic drug requires a new application costing $365,000 plus other fees, making 1,000 applications $365 million before additional expenses. Cuban said application costs, pods and FDA work all limit expansion, but predicted that daisy-chained pods could eventually make tablets and generics in Dallas for less than production with Cost Plus Drugs’ Indian partners, whom he said the company selects for quality rather than the lowest price.

Cuban also expects AI to help personalize medication using DNA and other variables, initially for over-the-counter drugs and eventually for prescriptions. He said designing an individual drug could approach zero cost in five, 10 or 20 years; DNA testing costs $99 or less, and a service might charge $99 upfront. Falling inference and personalization costs could let frontier AI companies include personalized medicine in subscriptions. He does not expect this to eliminate doctors: patients still need human care, clinicians must know which variables matter, and privacy concerns may deter people from sharing DNA. Wearable data, such as real-time blood-pressure readings, could help show how a person responds to treatment, diet or exercise, but turning individual circumstances into useful inputs may require human help.