The median U.S. late-stage round this year was exactly $100 million—what was once supergiant is now just humdrum, everyday dealmaking.
The venture capital market has undergone a structural shift so profound that Crunchbase retired the term "Supergiant Round" for deals above $100 million. The median U.S. late-stage round in 2026 hit exactly $100 million, double the $50 million median from 2020. The category that once denoted exceptional scale now describes typical late-stage financing. So far this year, investors have backed 250 rounds of $100 million or more, putting 2026 on track for year-over-year gains in deal count while capital raised already sits at record levels thanks to outlier rounds for OpenAI and Anthropic. This isn't market dysfunction—it's recalibration. The late 2010s saw initial traction with Uber and WeWork scaling ahead of IPOs, a 2021 bull-market peak, a pullback, and now resurgence driven by AI. What changed is where the money goes: fewer bets, larger checks, concentrated capital.
May delivered 29 new unicorns, but the composition tells the real story. The standout trend wasn't new foundation models but rather businesses helping enterprises deploy AI into operations. OpenAI and Anthropic each launched multibillion-dollar deployment ventures staffed with forward-deployed engineers. Infrastructure, autonomous software, and robotics startups dominated the cohort alongside healthcare, quantum, aerospace, financial services, manufacturing, and e-commerce plays. China led in robotics, Canada in quantum, Brazil contributed a legaltech unicorn, and OnlyFans raised its first external financing. Of the 29, seventeen were U.S.-based, four each from China and the UK, two from Canada, and one each from India and Brazil.
The Unicorn Board reached $9.9 trillion in value in May, with Anthropic moving ahead of OpenAI to become the second most valued private company after SpaceX, which listed in a record-setting IPO shortly after.
The headline rounds illustrate capital concentration at work. NinjaOne secured over $400 million in Series C extension at a $12.3 billion valuation after growing revenue over 70 percent in 2025 and posting first-quarter profit. Ramp closed $750 million led by Iconiq, GIC, and Ontario Teachers' Pension Plan at a $44 billion valuation. Three separate $500 million rounds went to Impulse Space for spacecraft propulsion, Supabase for open-source AI developer tools at a $10.5 billion valuation, and Flourish for brain-inspired AI models backed by Bezos, Lux Capital, and Google Ventures. Digital Asset raised $355 million in later-stage financing led by a16z crypto for blockchain technology aimed at financial institutions. TensorWave closed $350 million in Series B for AMD AI cloud infrastructure. These aren't moonshots—they're companies with traction, revenue, and paths to profitability.
Prometheus represents the logical extreme of this trend. The physical AI startup co-founded by Jeff Bezos and Vik Bajaj, former co-founder of Google's Verily, raised $12 billion at a $41 billion valuation from Bezos, JPMorgan Chase, Goldman Sachs, and BlackRock. This is Prometheus's second raise after an initial $6.2 billion late last year. The company is building what it calls an "artificial general engineer"—software to automate design and manufacturing of complex physical systems from jet engines to drug compounds. Bezos told CNBC the productivity gains will create "labor scarcity" where demand for human workers outpaces supply, predicting two-earner households becoming one-earner and overtime work declining. A large portion of capital will fund compute needs.
With 150 employees across San Francisco, London, and Zurich, Prometheus is keeping build specifics under wraps but represents the thesis that AI's next value creation wave happens in atoms, not just bits.
Mistral AI reportedly entered early discussions to raise approximately €3 billion at a €20 billion valuation, nearly double the €11.7 billion valuation from its September Series C, per Bloomberg citing anonymous sources. The French AI lab has positioned itself as a "sovereign" European alternative to American rivals, offering some foundational models with open weights while building closed models for programming, voice, and optical character recognition. Mistral is establishing a data center near Paris and has partnered with France's army, Luxembourg's government, and major European enterprises. Yet it has raised only about $4 billion to date per PitchBook—a fraction of OpenAI's $186 billion and Anthropic's $161.25 billion. The valuation gap reflects how far American labs have pulled ahead in revenue, model adoption, and enterprise demand.
Mistral did not return requests for comment on the rumored raise.
The vertical AI thesis is attracting serious capital even at early stages. Scotch raised $20 million in Series A led by VMG Partners with participation from First Round Capital, Lerer Hippeau, and Toba Capital for an AI-native operating system designed specifically for liquor store owners. The Denver-based startup provides point-of-sale hardware, custom software, payment processing, and back-office regulatory compliance tools. Customers range from single-register boutiques to enterprise stores running over a dozen lanes. Scotch reported greater than 500 percent year-over-year growth and surpassed $1 billion in processed payment volume. Co-founder and CEO Jake Bolling called the funding "a significant step-up" from the $10 million seed raised in September 2024.
Market research revealed the liquor store industry as fragmented with outdated point-of-sale technology—precisely the setup where AI-native vertical software can capture margin and deliver immediate ROI through compliance automation and inventory optimization.
European robotics startups are pulling in American-scale checks. Theker raised $85 million in what the Barcelona-based company calls "Europe's largest ever robotics Series A," led by CRV with backing from Samsung and Aglaé Ventures, the investment vehicle tied to LVMH chairman Bernard Arnault. Unlike humanoid robots designed around fixed forms, Theker's machines are built to be reconfigured—hands, arms, and overall form swap out depending on whether the task is sorting packages, packing clothing, or handling bottles in a warehouse. Inditex, Zara's parent company, signed on as an early backer, but co-founder Carla Gómez Cano told TechCrunch the broader goal is moving beyond retail into heavier manufacturing where manual task complexity and scale is greater.
Samsung is not yet a client but in advanced discussions, potentially becoming customer, supplier, and investor simultaneously—a trifecta offering both revenue and manufacturing credibility at scale.
The data confirms investor behavior has fundamentally changed, not broken. General Catalyst and Andreessen Horowitz led or co-led six rounds each in May, including Anduril's $5 billion Series H and Cognition's $1 billion Series D. The highest-spending lead investors were all co-lead backers in Anthropic's $50 billion Series H: Sequoia Capital, Altimeter Capital, Dragoneer, Greenoaks, Capital Group, Coatue, D1 Capital Partners, GIC, Iconiq Capital, and XN. More money than ever flows into jumbo-sized rounds, but as capital concentrates among hot names, deal volumes remain well below the 2021 peak. Direct sales are returning as annual contract values climb. Vertical AI and services deploying models—not just foundation labs—attract major dollars because they solve real-economy automation problems with measurable ROI.
The market recalibrated post-2021 excess, favoring capital efficiency and operational leverage over pure software multiples.
The semiconductor space illustrates the same dynamic. Investors poured around $10.7 billion into seed through pre-IPO rounds for semiconductor startups through May, putting 2026 on track to eclipse 2025 levels. MatX raised $500 million in Series B led by Jane Street for chips customized for large model needs. Ayar Labs secured $500 million in Series E led by Neuberger Berman for AI infrastructure focused on optics technology with strategic backing from AMD Ventures and Nvidia. Etched.ai reportedly raised $500 million at a $5 billion valuation led by Stripes for chips targeting AI superintelligence. Cerebras Systems went public in May raising over $5 billion, though shares have fallen about a third from initial closing prices.
The pattern holds: massive capital deployed into infrastructure enabling AI deployment at scale, with public market reception determining whether current private valuations prove durable or require correction.