AI startups are swallowing the venture capital market and the returns, so far, are looking good
Artificial intelligence startups are rewriting the rules in the venture capital market, and the 2025 numbers made that clearer than ever. But what really stands out is what came next: billion-dollar rounds that kept rolling in through 2026, IPO signals on the horizon, and a concentration of capital that is reshaping the very structure of the venture capital ecosystem.
According to the latest data from Carta, AI companies accounted for 41% of the $128 billion raised through venture capital over the past year. That is a historic record for annual share, and the most interesting part is that nobody was all that surprised. The market crowd already knew investor appetite for AI was through the roof, but seeing the official data confirm it at such a massive scale is a completely different story, especially when you stack those numbers up against what we saw in previous tech cycles.
To put it in better perspective, Carta itself revealed that 10% of startups captured half of all funding for the year. In other words, the money was not being spread evenly. It was being channeled surgically into a handful of companies that investors saw as ultra-high-conviction bets. That kind of concentration says a lot about where the market stands right now and about the expectations being placed on these companies.
And look, the investments did not slow down after that. Giants like OpenAI, Anthropic, and xAI kept raising billions in 2026, with valuations that left even the most seasoned industry veterans picking their jaws up off the floor. But it is not all sunshine and roses for everyone. The market is getting increasingly concentrated, with a few startups taking the biggest slice of the pie while the rest try to survive on the margins. The big question now is whether all this euphoria will translate into real returns, or whether we are watching the formation of something that might not end so well. 👀
What is behind the record-breaking numbers
When we say artificial intelligence startups absorbed nearly half of all venture capital deployed in 2025, it is easy to let the number slide by as just another tech market stat. But it is worth pausing for a second to understand the sheer scale of what this represents. We are talking about roughly $52 billion directed specifically to companies developing or applying AI as their core product, something with no historical precedent in any other innovation cycle, not even during the dot-com boom in the early 2000s or the explosive growth of mobile platforms in the decade that followed.
Carta, the platform that tracks cap table data and funding rounds for thousands of companies, provided a detailed breakdown of how those resources were distributed across quarters. What catches your attention is not just the volume but the consistency. Unlike other moments when specific sectors experience quick spikes of interest and then cool off, the flow of money into AI stayed hot throughout virtually every quarter of the year, with a particularly strong acceleration in the second half, when large Series B and C rounds started multiplying at an impressive pace.
Another factor that explains this investor behavior is the maturity of business models that started emerging. Early in the generative AI cycle, a lot of the money was being bet on research and infrastructure with not much concrete return on the horizon. Today, a growing share of startups already show recurring revenue, long-term enterprise contracts, and use cases that demonstrate measurable operational efficiency. That changes the conversation with venture capital funds, which shifted from viewing AI as a long-term bet to seeing it as an opportunity with return potential within shorter windows.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Peter Walker, head of insights at Carta, explained to TechCrunch that while rounds have become slightly harder to close, the capital amount per round has gone up. In practice, that means fewer bets but with bigger checks. And the reason is not that these startups have armies of employees. Quite the opposite — many operate with relatively lean teams. What makes the operation expensive is the cost of running AI models, which requires high-performance computing infrastructure with significant spending on GPUs, storage, and cloud processing.
The giants leading the charge
You cannot talk about the AI investment landscape without mentioning the companies that practically set the tone for the market in recent months. And when we look at early 2026, the numbers get even more jaw-dropping.
xAI, the artificial intelligence company founded by Elon Musk, kicked off the year by raising $20 billion in a Series E round in January. That infusion put the company on an entirely different level and sparked debates about whether the market is pricing in real potential or whether there is a significant hype component at play. The Grok model, integrated directly into the X platform, represents a different bet on how large language models reach end users, banking on organic access rather than relying solely on APIs and enterprise integrations.
Shortly after, in February, Anthropic raised $30 billion in its Series G round, hitting a valuation of $380 billion. With a stated focus on AI safety and alignment, the company attracted investors who wanted exposure to the sector without giving up a more responsible narrative around language model development. Its Claude model family gained real traction among developers and businesses throughout the period, and the partnership with Amazon brought in billions in funding that positioned Anthropic as one of the main alternatives to the OpenAI ecosystem, especially in the enterprise market where compliance and auditability matter just as much as performance.
But the one that truly shattered records was OpenAI. Also in February 2026, the company closed a $110 billion round, one of the largest private fundraises in history, pushing its valuation toward the symbolic $1 trillion mark. The product, the user base, and the enterprise agreements with companies like Microsoft remain the pillars supporting that valuation argument. Together, OpenAI and Anthropic represented a massive chunk of the $189 billion in global venture capital invested in February alone.
And there is more: all three companies — OpenAI, Anthropic, and xAI — have already signaled IPO plans for later this year, which has investors genuinely buzzing with the possibility of liquidity in a market where lucrative exits have been increasingly rare in recent years.
Concentration and the other side of the coin
With all the excitement around the numbers, there is a reality that deserves attention: the distribution of resources is far from democratic. The current state of the venture capital market is what experts call a K-shaped, or bifurcated, environment. On one side, capital concentrated in a few funds making big bets on a handful of companies. On the other, the rest of the ecosystem trying to compete for increasingly scarce resources.
While the big AI startups keep attracting massive rounds, a significant share of smaller companies is finding a market that is much tougher than the headlines suggest. Venture capital funds that used to place broader, more diversified bets are now concentrating their chips on companies with proven traction, teams with track records, and proprietary technology that is hard to replicate. This creates a brutally competitive environment for anyone trying to grow with more experimental theses or at intermediate stages of maturity.
The phenomenon creates a curious dynamic in the market. Seed-stage startups can raise with relative ease because the ticket sizes are smaller and the bet is more speculative. Companies that have already reached a reasonable level of scale also get access to large rounds. But the gap between those two points has become increasingly slippery, demanding a speed of execution from companies that is not always compatible with responsible technological development.
There is also a legitimate discussion around the valuations being practiced in this environment. When a company that has not yet reached breakeven receives a valuation of tens or hundreds of billions of dollars, the market starts wondering what metric anchors that number. Revenue multiples on projected earnings? Total addressable market size? Founding team reputation? In many cases, the honest answer is a combination of factors that include elements difficult to objectively quantify, which makes the due diligence process more challenging and, at times, less rigorous than it should be given the amounts involved.
Returns are looking good — but hold on
One of the most interesting data points from the Carta report relates to fund performance. Funds raised in 2023 and 2024, meaning after the launch of ChatGPT in late 2022, posted the highest internal rate of return (IRR) compared to funds raised between 2017 and 2020, which showed declining IRR. The report interprets this uptick as a positive indicator for funds that are positioned in the most promising startups of this AI cycle.
Peter Walker described the outlook as promising but made a point of adding context. He explained that younger funds can look like they are performing well on paper due to a natural effect of the investment cycle. If a fund invested in a Seed round and that company later raised a Series A at a higher valuation, the return on paper looks excellent over a short period. That pushes IRR up, but it does not necessarily mean money in investors’ pockets. Real returns only happen when there is a liquidity event, like an IPO or an acquisition.
Walker also pointed out that the portfolios of more recent funds are likely packed with AI-native startups, while vintage funds from 2020 and 2021 carry companies from earlier cycles that may not be riding the same valuation wave. That difference in portfolio composition explains part of the performance gap, but it does not tell the whole story.
The central question remains: will this early enthusiasm translate into real returns for investors through exits like major IPOs or billion-dollar acquisitions? Or are we just in the hype phase of a bubble that will eventually burst? Only time will tell.
What to expect going forward
The question hovering over all of this activity is whether the current cycle of investments in artificial intelligence has solid enough fundamentals to sustain the expectations being built, or whether there is a disconnect between the narrative of technological transformation and the reality of the returns that venture capital funds need to deliver to their limited partners. The history of technology markets shows that cycles of euphoria generally coexist with real fundamentals, but that multiple compression and expectation resets tend to arrive at some point, especially when the pace of adoption does not keep up with the pace of fundraising.
What sets this moment apart from other cycles, according to analysts who follow the sector closely, is the speed at which commercial applications are taking hold. Unlike the dot-com bubble, where most companies had no clear business model, the AI startups raising massive volumes in 2025 and 2026 largely present real contracts, paying customers, and use cases that solve concrete problems within large organizations. Process automation, content generation at scale, predictive analytics, and copilots for knowledge workers are examples of categories with proven demand and fast-growing revenue.
The potential IPOs of OpenAI, Anthropic, and xAI expected later this year will serve as a kind of reality check for the entire ecosystem. If these public offerings confirm the valuations from the private market, the cycle gains even more momentum and validates the thesis that generative AI is, in fact, a transformative platform capable of generating returns proportional to the capital invested. If there is a correction, the impact will be felt across the board, from the mega funds that led the rounds all the way down to early-stage accelerators that built their portfolios around the AI narrative.
Until then, the flow of venture capital into AI should stay heated, but with growing selectivity that will more clearly separate companies with a solid value proposition from those that rode the wave without building something genuinely differentiated. The market is going through a natural maturation phase, where the excitement of the early days starts giving way to more careful analysis of unit economics, real scalability, and the ability to generate sustainable revenue over the long run.
For anyone following the sector, the message from Carta’s data is clear: AI is no longer a side bet in the venture capital portfolio. It is the center of gravity. And like any center of gravity, it will keep pulling more mass toward it, for better and for worse. 🚀
