AI funding hit $189B in a single month, taking 83% of all global VC capital
February 2026 saw AI companies raise $189B, more than all of 2025 combined. Here's what the numbers actually mean.
Whether you're looking for an angel investor, a growth advisor, or just want to connect — I'm always open to great ideas.
Get in TouchAI, startups & growth insights. No spam.
TL;DR: In February 2026, AI companies raised an estimated $189 billion in venture capital, a figure larger than all AI funding in 2025 combined. OpenAI's $40B round and Anthropic's $30B raise anchor the numbers, but infrastructure bets on CoreWeave, Lambda, and Crusoe pushed the total into truly unusual territory. Whether this is rational capital allocation or a replay of 1999 depends almost entirely on whether current revenue trajectories hold.
The total AI venture funding in all of 2025 was approximately $130 billion, according to Crunchbase data. February 2026 alone exceeded that figure. This is not a rounding error or a matter of disputed methodology. It is a real structural shift in where institutional capital is flowing.
The $189 billion estimate comes from aggregating announced rounds across foundation model labs, AI infrastructure companies, and applied AI software firms. Roughly 83% of all global venture capital deployed in February went to AI companies. The remaining 17% covered every other sector combined.
The comparison table below captures the scale of the shift:
| Metric | 2025 full year | February 2026 alone |
|---|---|---|
| Total AI VC raised | ~$130B | ~$189B |
| Share of global VC | ~45-50% | ~83% |
| Largest single round | ~$6B (xAI) | $40B (OpenAI) |
| Number of $1B+ rounds | ~12 | ~8 |
| ✓ Exceeds prior year total | ✗ | ✓ |
| ✓ Majority of all global VC | ✗ | ✓ |
These numbers are not evenly distributed. Two companies account for $70 billion of the $189 billion total. That concentration matters enormously for how you interpret the headline.
OpenAI closed a $40 billion funding round in early 2026, led by SoftBank, with participation from existing investors. The round values the company at $300 billion post-money. No private company has ever raised this much in a single round.
The context matters. OpenAI's annualized recurring revenue had reached approximately $12.7 billion by early 2026, up from roughly $3.4 billion at the start of 2024. That is real revenue from real customers paying for ChatGPT subscriptions, API access, and enterprise contracts. The valuation implies roughly 24x forward revenue, which is aggressive but not insane for a company growing at this pace.
The capital is intended to fund compute expansion, research into more capable models, and international expansion. OpenAI is also in the middle of a corporate restructuring from capped-profit to a more conventional for-profit structure. The new round is partly contingent on that restructuring completing. If it does not complete, terms reportedly allow investors to exit at a stepped-up price. That is an unusual provision and worth watching.
Sam Altman has been clear that the capital requirements for frontier AI are larger than any single company can self-fund. The $40B round is a statement that the bet on artificial general intelligence requires a war chest comparable to building a semiconductor fab.
Anthropic has raised over $30 billion across multiple rounds, with major commitments from Amazon and Google. The Amazon relationship is particularly significant. AWS committed up to $4 billion in an earlier tranche, and Anthropic agreed to use AWS Trainium and Inferentia chips as a preferred compute platform.
Anthropic's revenue is smaller than OpenAI's, running at roughly $2 billion ARR, but the company has carved out a defensible position in the enterprise market. Claude is the preferred model for many financial services and legal technology deployments, where companies have specific concerns about data privacy and output reliability.
The company's "safety-first" positioning used to read as a marketing differentiator. It has since become a genuine competitive advantage as regulators in the EU and UK increasingly require documentation of safety practices before enterprise deployments.
Dario Amodei, Anthropic's CEO, has argued publicly that safety research and capability research are complementary, not opposed. That framing has helped Anthropic raise capital from investors who might otherwise be skeptical of spending billions to build systems with uncertain commercial timelines.
The foundation model labs get the headlines, but a significant portion of the $189 billion total went to AI infrastructure companies. CoreWeave, the GPU cloud provider, raised billions more in its pre-IPO financing rounds and is preparing for a public offering that could value it above $35 billion. Lambda Labs and Crusoe Energy, two other compute-focused companies, also raised substantial rounds.
This matters for a few reasons. First, the infrastructure layer is less winner-take-all than the model layer. Multiple GPU cloud providers can coexist. Second, these companies have actual hardware, actual customers, and more predictable revenue models than the foundation labs. Third, infrastructure investments are a bet that AI compute demand will remain high regardless of which model wins.
Crusoe Energy's angle is particularly interesting. The company sources compute from stranded natural gas and renewable energy, positioning itself as the "green compute" option for AI training. In an environment where data center power consumption is becoming a political issue, that positioning has value beyond the spreadsheet.
The combined infrastructure investment is a bet that the physical layer underneath AI is a durable business. That is a safer bet than it looks.
Several other significant rounds closed in early 2026. xAI, Elon Musk's AI company, raised $6 billion at a valuation of $50 billion. The company's Grok model has been integrated into X (formerly Twitter), giving it distribution that most AI companies would pay heavily for.
Wayve, the UK-based autonomous driving company, raised $1.5 billion in a round led by SoftBank. Wayve's approach differs from Waymo's in that it uses a single end-to-end neural network rather than a modular system. The bet is that a generalist driving model trained on internet-scale data will outperform hand-engineered systems over time.
Figure AI raised $675 million for humanoid robotics. The company has a partnership with BMW for factory deployment. General-purpose humanoid robots are still early, but the investment thesis is that a sufficiently capable foundation model combined with a capable body creates a general-purpose labor substitute. That is a very large market if the thesis is right.
These rounds collectively represent billions more but point to a broader pattern: capital is following companies that have either real revenue, real distribution, or a plausible path to replacing expensive human labor.
When AI companies take 83% of all venture capital deployed globally in a single month, the obvious question is: what happened to everyone else? The answer is that non-AI funding did not collapse. The total pool of venture capital deployed globally in February 2026 was very large, and AI captured an outsized share of it.
But the concentration is real and has real consequences. Late-stage biotech, climate tech, fintech, and enterprise software companies are all competing for a much smaller share of investor attention and capital. Several well-regarded venture funds have publicly shifted their focus almost entirely to AI. The opportunity cost is paid by founders in other sectors who cannot raise at reasonable terms.
The more subtle consequence is within AI itself. The concentration is not evenly spread across AI companies. It is highly concentrated among a small number of frontier labs and infrastructure providers. The $189 billion total could give the impression that AI startups generally are well-funded. Most are not.
Application-layer AI companies, the ones building AI-powered products on top of the foundation models, are in a complicated position. They can move fast and generate revenue, but their moats are thin if the foundation model providers decide to compete directly. This dynamic is suppressing application-layer valuations even as foundation model valuations soar.
The bulls have a real case. OpenAI's revenue growth from $3.4B to $12.7B ARR in roughly two years is one of the fastest revenue ramps in enterprise software history. Anthropic's $2B ARR, while smaller, represents genuine enterprise adoption. These are not zero-revenue companies at absurd valuations.
The bears also have a real case. The total capital raised by AI companies in the past 24 months is now measured in the hundreds of billions. Even at OpenAI's growth rate, the time required to generate returns on that capital is long. The implicit assumption in current valuations is that AI becomes a multi-trillion-dollar market. That may be correct. It is not guaranteed.
The most honest reading of the data is that a small number of AI companies are generating real, fast-growing revenue, and investors are using that fact to justify much larger bets on companies that have not yet matched that performance. This pattern is familiar from previous technology cycles.
One data point worth tracking: OpenAI's compute costs are enormous. The company reportedly spends more on compute than it earns in revenue, even at $12.7B ARR. That means profitability requires either significant revenue growth, significant cost reduction, or both. The $40B raise buys time for those conditions to emerge.
The 1999 internet bubble and the current AI cycle share structural similarities. Both involve a genuinely new general-purpose technology. Both have attracted capital far in excess of current earnings. Both have created companies with enormous valuations and plausible but uncertain paths to profitability.
The differences are meaningful. In 1999, many of the largest dot-com companies had negligible revenue. OpenAI's $12.7B ARR and rapid growth would have been extraordinary in any era. The underlying technology in 2026 is also more demonstrably useful. AI is being deployed in real production environments at scale, not just as a vision of what the internet might become.
The more relevant comparison may not be the dot-com bubble but the semiconductor investment cycle of the 1990s. Enormous capital flowed into chip companies, some of which failed and many of which built durable businesses. The survivors went on to generate returns commensurate with the investment. The question is whether AI follows the semiconductor pattern or the dot-com pattern.
The answer is probably both. Some of the current AI investments will generate extraordinary returns. Many will not. The challenge for investors is that it is very early to distinguish the winners.
The flip side of $189 billion going to AI is the sectors and geographies that are not getting capital. Enterprise software without an AI story is raising at compressed multiples. Consumer fintech has largely stalled. Climate tech is competing for a fraction of what it raised two years ago.
Within AI, the companies not getting funded are instructive. Application-layer startups in crowded categories, like AI writing, AI image generation, and general-purpose AI assistants, are struggling to raise at meaningful valuations. The argument from investors is that these categories will be commoditized by the foundation models themselves. That argument has some merit.
Companies building narrow, domain-specific AI tools for highly regulated industries, like healthcare, legal, and financial services, are raising, but at lower multiples than the foundation labs. Investors want the big bet, not the niche play.
Early-stage AI research companies without clear commercial paths are also finding it hard to raise outside of a small number of specialized investors. The market has moved toward companies with existing revenue or a very short path to revenue.
Every investment cycle ends. The question for AI is not whether capital will eventually become scarcer but what happens to the industry when it does.
The most likely scenario is that the top two or three foundation model labs continue to receive capital because they have demonstrated revenue and defensible technical positions. The rest of the market consolidates aggressively. Several well-funded AI startups that raised at high valuations will merge or shut down. The infrastructure layer, with its tangible assets, will weather the transition better than pure software plays.
Application-layer companies built on top of a single foundation model will be at risk if that model's pricing changes or if the model provider decides to compete directly. Diversification across model providers is a rational hedge that most application companies are not yet prioritizing.
The scenario where this ends badly is one where compute costs do not fall as fast as expected, revenue growth slows, and a handful of high-profile AI companies fail to hit their implied growth targets. That would trigger a revaluation across the sector and a significant reduction in available capital. It would also create a substantial opportunity to hire talented people and acquire distressed assets at rational prices.
A few near-term indicators will tell a lot about the durability of the current funding environment.
OpenAI's corporate restructuring is the most important single variable. The $40B round's completion depends on it. If the restructuring hits legal or regulatory obstacles, the round could unwind or restructure, which would affect market sentiment significantly.
CoreWeave's IPO, expected in mid-2026, will be a real-world test of whether public markets share private market enthusiasm for AI infrastructure. The IPO pricing will be a benchmark for every AI infrastructure company that follows.
Revenue growth at the major labs matters more than valuation in the next 90 days. If OpenAI's growth rate decelerates materially from the current trajectory, the implied valuation becomes harder to defend. Quarterly revenue updates from Anthropic will also be closely watched by investors trying to understand whether the enterprise AI market is as large as the funding numbers imply.
Finally, regulatory developments in the EU and the US will shape which applications are actually deployable at scale. Regulatory clarity is a prerequisite for enterprise adoption in several large categories. Movement toward clarity is bullish. Unexpected restrictions are a material risk.
An estimated $189 billion, based on aggregate data from Crunchbase and PitchBook. This is an estimate, as not all rounds are immediately disclosed, and some figures include previously committed capital being deployed in new tranches.
Total AI venture funding in 2025 was approximately $130 billion across the full year. February 2026 alone exceeded that total. This is a stark acceleration even accounting for the compound growth rate the sector had already demonstrated.
OpenAI raised $40 billion in a round led by SoftBank. The round values the company at approximately $300 billion post-money, making it the highest-valued private company in history at the time of close.
OpenAI's annualized recurring revenue reached approximately $12.7 billion in early 2026. The company has grown revenue rapidly from around $3.4 billion ARR in early 2024, representing one of the fastest enterprise software growth trajectories on record.
Anthropic has raised over $30 billion across multiple funding rounds, with Amazon and Google as major strategic investors. Amazon committed up to $4 billion in an earlier round tied to a cloud partnership agreement.
Anthropic is running at approximately $2 billion ARR. The company has a strong position in enterprise segments, including financial services and legal technology, where data privacy and output reliability requirements are stringent.
Other notable rounds include xAI at $6 billion, Wayve at $1.5 billion, and Figure AI at $675 million. CoreWeave also raised significant pre-IPO capital ahead of a planned public offering.
It means AI companies captured roughly 83 cents of every dollar deployed in global venture capital during February 2026. All other sectors, including biotech, climate, fintech, and enterprise software, shared the remaining 17%.
It depends on which companies you are looking at. OpenAI's valuation is aggressive but grounded in real, fast-growing revenue. Many other AI companies lack comparable revenue foundations. The honest answer is that some current investments will prove to be rational and many will not.
The key difference is revenue. Many of the largest dot-com companies had near-zero revenue. OpenAI's $12.7B ARR and Anthropic's $2B ARR are real. The risk is that capital is being deployed at scale to companies that have not yet demonstrated comparable commercial traction.
CoreWeave is a GPU cloud provider that rents compute infrastructure to AI companies. It is one of the largest beneficiaries of the AI infrastructure build-out. Its planned IPO will be a public market test of how investors value AI infrastructure businesses.
Investors worry that application-layer AI categories will be commoditized by foundation model providers who can offer similar features directly. The moats in application AI are thinner than in foundation models or infrastructure.
SoftBank led the $40 billion round, committing the largest single check. SoftBank's Vision Fund has a history of large concentrated bets in technology. This is consistent with that pattern but at a scale that exceeds any previous Vision Fund investment.
A material deceleration in revenue growth at the major labs would trigger revaluations across the sector and reduce available capital. This would likely accelerate consolidation among application-layer companies and pressure infrastructure valuations.
Several prominent venture firms have shifted their portfolios heavily toward AI. Others are doubling down on non-AI sectors where competition for deals is lower. The firms that continue to invest broadly in non-AI sectors have publicly noted that some of the best investment opportunities are in less crowded categories.
xAI raised $6 billion to fund development of its Grok model and expand compute capacity. The company has distribution advantages through its integration with X, which has hundreds of millions of users. The round values xAI at $50 billion.
Wayve is a UK-based autonomous driving company using an end-to-end neural network approach. SoftBank led its $1.5 billion round. The bet is that a generalist AI approach to driving will outperform traditional modular autonomous vehicle systems over time.
Yes. Healthcare, defense technology, and climate infrastructure companies with strong revenue and clear regulatory paths are raising, though at lower multiples than two years ago. The challenge is that the talent and capital market is tilted heavily toward AI.
EU AI Act compliance requirements, potential US executive orders on foundation model development, and antitrust scrutiny of the major tech companies' investments in AI labs all represent material regulatory risks. Clarity on these issues is a prerequisite for certain large-scale enterprise deployments.
Crunchbase, PitchBook, and The Information track funding rounds with varying levels of detail. Bloomberg and Financial Times cover the major rounds with additional context on terms and investor motivations.
Trump's AI executive order preempts state AI laws as FTC comment deadline hits. What OpenAI, Anthropic, and 36 AGs are doing about it.
Anthropic Pentagon ban explained: DOD named the AI lab a supply chain risk. Microsoft filed an amicus brief. Here's what it means for AI procurement.
OpenAI has acquired Promptfoo, the open-source AI security testing platform used by 130,000 developers and 25% of Fortune 500 companies, to harden its agentic AI systems against prompt injection, jailbreaks, and adversarial attacks.