- HEADLINE & INTRO
Venture capital just had its wildest month on record – and it was essentially a three‑company show. February’s $189 billion in global VC funding looks, at first glance, like a spectacular comeback for startup investment. In reality, it’s a snapshot of how distorted the market has become when AI infrastructure and robotaxis sit at the center of the tech universe.
In this column, we’ll unpack what these eye‑watering rounds for OpenAI, Anthropic and Waymo really mean: for competition, for smaller startups that saw almost none of that money, and for a world where a “startup” can be valued like Apple but regulated like… nothing at all.
- THE NEWS IN BRIEF
According to TechCrunch, citing fresh Crunchbase data, global venture capital hit a record $189 billion in February 2026. Around 90% of that – roughly $171 billion – went to AI‑related startups.
The surge was driven by three colossal late‑stage deals. OpenAI reportedly raised about $110 billion at a valuation of $730 billion, in what Crunchbase categorises as one of the largest private funding rounds in history. Anthropic followed with a $30 billion Series G at a $380 billion valuation. Waymo, Alphabet’s self‑driving unit turned standalone company, secured $16 billion at a $126 billion valuation.
Together, these three companies attracted roughly $156 billion, or 83% of all VC dollars deployed globally in February. Crunchbase notes that the amount invested in this trio alone equals about one‑third of the total global venture spending recorded for the whole of 2025.
- WHY THIS MATTERS
These numbers are not just big; they are system‑reshaping.
First, they confirm that venture capital is no longer primarily about seeding thousands of risky bets. It is increasingly about underwriting private mega‑infrastructure: frontier AI models and autonomous mobility platforms that require gigantic compute, data and capex budgets. Capital itself becomes a competitive moat – if you do not raise tens of billions, you simply cannot play at the top end of this market.
Second, this concentration is brutal for everyone outside the inner circle. On paper, February was a record month for founders. In practice, the vast majority of startups saw none of that capital. LPs and large funds that piled into these rounds will have less room and less appetite for smaller, riskier tickets. Expect even more pressure on early‑stage funds and on founders in less fashionable sectors – think B2B SaaS, climate, hardware – to do more with less.
Third, these valuations push the line between “startup” and “systemically important platform.” When a private AI company is valued in the same league as the largest public tech firms, but without the same disclosure, scrutiny or shareholder discipline, regulators and governments cannot ignore it. Access to compute, training data and model APIs will increasingly look like critical infrastructure – especially once these systems are embedded into finance, healthcare, defence and public administration.
The winners today are the three companies and their backers. The potential losers are would‑be competitors, smaller ecosystems starved of late‑stage capital, and ultimately users and regulators who find themselves negotiating with a tiny club of AI gatekeepers.
- THE BIGGER PICTURE
This wave of mega‑rounds fits several longer‑running trends.
We have seen capital concentration before: the SoftBank Vision Fund era around 2017–2019, the 2021 boom in quick commerce and fintech, and the late‑stage feeding frenzy around companies like WeWork. Those cycles ended with painful down‑rounds and write‑offs when public markets refused to endorse the private valuations.
There are key differences this time. Foundation models and autonomous driving are extraordinarily capital‑intensive and sit at the base layer of many future applications. Unlike food delivery or flexible offices, they can plausibly justify enormous infrastructure investment: data centers, chips, specialised hardware, global fleets. That does not eliminate bubble risk, but it makes the story more than just financial engineering.
The flipside is that such capital intensity locks in a power law. Only a handful of global players – OpenAI, Anthropic, maybe a couple of Chinese giants, plus a few autonomous driving leaders like Waymo – can realistically raise enough money to stay on the frontier. Everyone else will be forced into narrower niches: domain‑specific models, vertical applications, or open‑source ecosystems that piggyback on existing infrastructure.
Competitively, this cements an AI stack where a few US‑based private platforms sit between hyperscale clouds above and millions of application developers below. It also accelerates convergence between venture capital, sovereign wealth and strategic investment from big tech. When rounds hit $100 billion, the investor base is closer to macro‑finance than to traditional Sand Hill Road.
In short: this month’s deals are not an outlier. They are a preview of an AI age in which “venture‑backed startup” can mean a private behemoth bigger than most stock‑listed companies.
- THE EUROPEAN / REGIONAL ANGLE
For Europe, these deals underline a hard truth: the frontier of general‑purpose AI is being set elsewhere, but its consequences will be felt very directly here.
No European AI company is raising anything close to $110 billion. Even the most prominent regional players – from German model builders and translation specialists to French and UK labs – operate two orders of magnitude below that capital scale. That makes it almost impossible to compete head‑on with US‑based foundation models on raw compute and model size.
But Europe does hold leverage in two other ways. First, regulation. The EU AI Act, together with the Digital Markets Act and existing competition rules, gives Brussels significant power to dictate how these mega‑funded players can operate, bundle services and use data in the single market. If access to frontier models becomes essential infrastructure, the Commission will be tempted to treat them a bit like cloud providers or operating systems: subject to strict obligations, interoperability requirements and perhaps data‑sharing.
Second, Europe has room to define alternative paths: open‑source models, specialised vertical AI, robotics, energy‑efficient architectures and sovereign cloud. For ecosystems from Berlin to Paris, Barcelona to Ljubljana and Zagreb, the opportunity is to build on top of, or around, these mega‑platforms rather than to mimic them.
The risk is clear: if European policymakers over‑rotate on regulation without pairing it with serious industrial strategy and funding, local founders may end up as mere API customers of a tiny US oligopoly.
- LOOKING AHEAD
Where does this go next?
In the near term, more mega‑rounds are likely. Other capital‑hungry domains – AI‑enabled defence tech, space, advanced chips – are already attracting outsized interest. The sidebar in TechCrunch about defence unicorn Anduril hunting a massive valuation is not a coincidence; investors are chasing categories where scale and capital intensity create defensible moats.
For founders, the market will polarise. A microscopic club of companies will raise tens of billions to own foundational infrastructure. A much larger group will focus on leaner models, data network effects in specific industries and capital‑efficient businesses. The middle – $500 million to multi‑billion late‑stage growth stories without clear moats – will be a very uncomfortable place to sit.
Key things to watch over the next 12–24 months:
- Exit paths: Can any of these mega‑funded AI players go public at or above current valuations without crashing the market?
- Regulatory posture: How quickly do the EU, US and others move from generic AI rules to company‑specific remedies and investigations?
- Dependency risk: How many banks, hospitals, public agencies and startups build deep dependence on one or two model providers before there are credible alternatives?
If public markets balk or regulators step in aggressively, today’s exuberance could look like the top of the cycle. If not, we may be seeing the birth of a new asset class: privately held, quasi‑sovereign AI infrastructure companies.
- THE BOTTOM LINE
February’s record VC numbers do not signal a broad startup renaissance; they mark the arrival of a hyper‑concentrated, AI‑centric capital regime. Three companies have just soaked up a year’s worth of funding for an entire generation of startups.
The question now is whether policymakers, founders and investors are willing to accept a future where the foundations of intelligence and mobility are controlled by a handful of privately financed giants – or whether they will use this moment to build alternatives, before it is too late.
