This is a cool moment for the era of mega-rounds in artificial intelligence. A fresh tally reveals that 49 U.S.-headquartered AI startups have raised just over $100 million in capital so far this year, what with six new companies joining the list in recent weeks, which means we’ve had a banner pace on mega-deals, if not surpassing last year’s torrid activity, underscoring investor confidence that a small clutch of AI platforms and category leaders will vacuum up outsized capital as adoption accelerates.
It’s not just the size of the club, merely, that stands out in 2025 but also the velocity of it. Multiple early-stage companies closed more than one nine-figure round during the year as they worked to secure compute, hire scarce talent and lock in enterprise distribution. That pattern — of more repeat mega-rounds, and earlier in the life cycles of companies — is a dramatic departure from previous venture cycles, recent analyses from PitchBook and CB Insights suggest.
What the 49 signal about AI funding in 2025
The 49-member class represents where value is pooling: foundation model labs and agentic platforms; data and training infrastructure; vertical AI in healthcare, finance, industrials and logistics; AI-native cybersecurity; robotics and autonomy; and chip design and acceleration software. The through-line is capital intensity. Training frontier models, orchestrating multimodal agents, and deploying AI at scale in safety-critical environments all require massive investment in GPUs, data and compliance.
Enterprise is starting to catch up with the hype cycle. Surveys tracked by Stanford’s AI Index and the NVCA indicate that wide-ranging tests are becoming paid rollouts, particularly in customer service, software development and back-office workflows. Alongside this, hospital systems are seeing double-digit percentage drops in their documentation time using ambient clinical AI, and financial institutions are reducing fraud false-positive rates with AI and speeding up underwriting reviews at the same time. Those tangible ROI stories are a major reason why growth funds and strategics leaned in with nine-figure checks this year.
How we counted the 49 and defined qualifying rounds
The count includes U.S.-based companies that are working on AI products, or whose key product is powered by AI models, data or agentic systems. It includes announced or disclosed primary equity rounds from $100 million and up this year, and some select structured financings and debt facilities when they were specifically tied to compute procurement. It does not include acquisitions, PIPEs linked to public listings or corporate in-house projects. The tally is based on SEC filings, company announcements and data from PitchBook, CB Insights and the NVCA as well as reporting.
Notably, each company is counted once even if it raised more than one round of $100 million or more during the same year. That distinction becomes salient in 2025, when there have been several breakout startups who layered a growth round onto a compute facility or extended their Series within months in order to lock down GPU supply.
Where the mega-rounds flowed across AI sectors
Foundation models and agentic platforms took home the biggest portion of mega-round dollars. These companies are moving from text to vision, audio and action — stitching the tools together so that software can perform multiple steps. That appetite for capital is linked straight to training runs and inference scale, where ramp access to the newest accelerators and top-shelf proprietary data sets continues as the winning advantage.
Another main stop was Vertical AI. In healthcare, ambient documentation and clinical decision support drew large rounds as both systems began to hook into EHRs, while also navigating FDA-facing workflows. In financial services, anti-fraud models and AI underwriting and risk copilots drew growth funding from banks and VCs focused more on fintech. Industrial AI received new capital for inspection, predictive maintenance and warehouse automation that marries computer vision with robotics — a place where AI translates into fewer defects, faster cycle times and safer work.
On the infrastructure front, there were data tooling startups purifying, labeling, evaluating and securing training sets included in the $100 million club as well as vector databases, retrieval systems and model observability providers. There were chips and systems software, of course, and companies that are working to better tune compiler stacks or memory throughput in the hopes of wringing a little more performance out of existing hardware — a more real-world hedge as top-shelf GPUs stay supply-constrained.
Who wrote the checks behind these AI mega-rounds
Strategic investors played a large part. Hyperscale and leading semiconductor companies either took part in rounds or led some of them, frequently combining equity investments with multi-year cloud credits, model distribution and compute commitments. That “money plus go-to-market” package helped startups land enterprise deals more quickly and helped to get gross compute costs down at scale.
On the venture side, multistage firms with strong AI theses — in addition to crossover and sovereign investors — led a number of these rounds. Notably, more boards approved so-called structured financings which combine primary capital and secondary liquidity for early employees, a development focused on retention in an ultra-competitive talent market.
Geography and stage dynamics in this year’s deals
The Bay Area maintained gravitational pull as the 49 pivoted, with New York, Boston, Seattle and Pittsburgh becoming meaningful nodes. Most mega-round recipients were at a Series B or C, while 2025 also saw some unusually large financings for early-stage teams with actual model results or unique domain datasets. The phenomenon of repeating mega-rounds — once a rare outlier — is now table stakes in a market where companies pre-purchase compute or scale their inference capacity to satisfy enterprise SLAs.
What to watch next as AI funding cycles evolve
The quality of the runway and revenue will determine what comes next. With GPU prices lingering and competition heating up, investors are now digging into gross margins post-cloud discounting, unit economics of inference-heavy product lines as well as the sustainability of data advantages. Look for more deals that trade lower COGS for platform exclusivity, ongoing use of debt to finance compute and an increase in M&A as adjacent players combine and offer stacked solutions.
For now, the message is clear: capital is coalescing around AI startups that can translate model horsepower into defensible distribution and measurable ROI. It’s that handful of 49 businesses that crossed the $100 million threshold this year that will be the ones to watch as we move beyond experimentation and into scaled deployment.
