AI startups are proving they can build real businesses, and Forbes' 2026 lists show exactly where the money is going

Anthropic's revenue run rate crossed $30 billion in early April 2026. OpenAI was already reporting more than $25 billion in annualized revenue as of February. Those two figures, drawn from Forbes' eighth annual AI 50 list published in April, do more to describe the current state of enterprise AI than any forecast: the foundation model market is no longer theoretical, and the capital structures around it are enormous.

The AI 50 and the separately published Next Billion-Dollar Startups list, released by Forbes on July 28, together form a useful operational map for procurement and technology leaders trying to understand which AI vendors are capitalizing at scale and which pre-unicorn companies are closing in on that threshold. The two lists span the full maturity spectrum, from companies running at tens of billions in annual revenue to sub-$250 million-valued startups that have barely begun to scale.

Two companies, 80% of the capital

The concentration of capital on the AI 50 is striking. According to Forbes, the 50 companies on this year's list have collectively raised $305.6 billion in venture funding. OpenAI and Anthropic account for $242.6 billion of that total, roughly 80 percent. For enterprise buyers evaluating vendor stability and roadmap continuity, that level of concentration matters: the two companies are not just the largest AI vendors by revenue, they are effectively subsidizing the competitive dynamics of the entire private AI market.

Both companies are reportedly moving toward public offerings. Their scale is reshaping adjacent markets, too. Forbes notes that Anthropic's Claude Code and OpenAI's Codex are pushing directly into the AI coding segment, where competitors like Cursor, valued at $29.3 billion, must continuously differentiate. For enterprise software and IT operations teams, that competitive pressure is accelerating the pace of product updates and pricing changes across coding assistant tools.

Beyond the giants: where the next tier is building

The more operationally instructive part of the AI 50 is what the non-Anthropic, non-OpenAI companies are doing. San Francisco-based Physical Intelligence has raised $1 billion to train foundational models for robots, using data collected from human teleoperators working in realistic environments. French startup Mistral is selling open-weight models to large corporations including Cisco and to European government agencies, where its local regulatory positioning is a differentiator. Coding-focused Cognition, valued at $10 billion, acquired the remaining assets of Windsurf after Google paid $2.4 billion to license Windsurf's technology and bring on its cofounders.

Several newcomers on the AI 50 carry metrics that enterprise technology evaluators should note. Gamma, an AI presentation builder valued at $2.1 billion, crossed $100 million in annualized revenue with just 50 employees. Rogo, based in New York, reports that roughly 25,000 bankers and investors use its AI software for financial analysis. Chai Discovery, a two-year-old startup valued at $1.3 billion, is applying AI to drug development and candidate identification. And Reflection, valued at $8 billion, is building open-source models specifically positioned as an alternative to Chinese AI companies including DeepSeek.

The list also reflects notable leadership. Former OpenAI CTO Mira Murati's Thinking Machines Lab has raised $2 billion, and Stanford professor Fei-Fei Li's World Labs, focused on spatial intelligence, has secured more than $1 billion in funding. Both are among four female-led companies on this year's AI 50, according to Forbes.

Pre-unicorn signals: healthcare and infrastructure lead the next wave

Forbes' Next Billion-Dollar Startups list, built in partnership with TrueBridge Capital Partners over 12 years, has a verified conversion rate: 60% of its 275 alumni have become unicorns, including Duolingo and DoorDash. Another 60 were acquired or merged. Nearly half of last year's picks have already crossed $1 billion in valuation. This year's cohort is notable because, according to Forbes, nearly all of the companies use AI in some form, spanning bone marrow treatment, biosecurity, cybersecurity, and creative design.

Two 2026 entries illustrate the operational breadth of where AI is being applied below the unicorn line. Abby Care, valued at $225 million and backed by Sequoia Capital, Thrive Capital, and Khosla Ventures, has built an AI-assisted platform that trains family members of disabled or elderly patients to become paid Medicaid caregivers. The company's app consolidates timesheets, clinical charting, and an AI-powered assistant for caregivers. It has raised $45 million and is targeting the same platform-marketplace model that Uber applied to transportation.

American Terawatt, a San Francisco-based infrastructure startup valued at $350 million, is addressing a more immediate data center problem. The company, which has raised $52 million from Altimeter Capital and Founders Fund, is building direct-current power grids specifically to serve AI data centers. The case for DC infrastructure is straightforward: electronics require stable DC power, but most grid power arrives as alternating current and must be converted. Energy is lost in that conversion. American Terawatt's approach eliminates it by delivering DC directly through buried transmission lines. CEO Anton Troynikov previously founded Chroma, the vector database used widely by AI applications for memory and retrieval.

What the lists signal for enterprise technology teams

Read together, the two Forbes lists describe a market that has moved past proof-of-concept. The AI 50 shows that the largest vendors are generating real revenue at scale and are competing directly with each other in categories like coding, legal, and financial analysis. That means enterprise buyers can now negotiate on the basis of demonstrated capability and commercial track record, not just roadmap promises.

The Next Billion-Dollar Startups list shows where the next round of specialized vendors is forming. Healthcare operations, data center power infrastructure, and sector-specific AI platforms are all represented in this year's cohort. For procurement and IT leaders building three-to-five-year vendor strategies, these are the companies worth evaluating before their valuations reflect broader market recognition. Forbes' 60% historical conversion rate suggests the signal is credible. The next step is identifying which of this year's picks are already inside enterprise pilots.