AI startups are eating the venture industry and the returns, so far, are good

When you look at last year's data, the reality is almost shocking: 41% of all venture capital funds — exactly 52 billion dollars out of 128 billion — went to startups focused on artificial intelligence. This is a record percentage that shows how drastically the landscape of technology investments has shifted. But that's only half the story. The actual concentration of capital is even more extreme: just 10% of AI startups consumed half of all the money allocated to the entire industry. This means that while thousands of companies fight for every dollar, a handful of giants — Anthropic, OpenAI, xAI — absorb capital at a pace that traditional venture capitalists might once have considered impossible.

This is no longer a trend. This is a fundamental transformation of the venture capital ecosystem that is changing the rules of the game for investors, entrepreneurs, and the entire technology industry. The question is no longer "is AI an important part of venture investments?", but rather "does anything else still matter?" The numbers speak for themselves, but their implications run much deeper than simple capital allocation statistics.

When 10% of startups consume 50% of the money

The concentration of capital in the AI industry has reached levels that deserve deeper analysis. This is not about natural distribution, where a few leaders receive more than the rest — this is something completely different. When just a few companies receive half of all available funds, the remaining 90% of startups must compete for the remaining 50%. This creates an extremely uneven playing field, where the difference between winner and loser is not a matter of better strategy or product, but access to capital.

Anthropic, OpenAI, and xAI are no longer startups in the traditional sense — they are quasi-corporate entities with valuations that rival many Fortune 500 companies. OpenAI recently raised 110 billion dollars in a single funding round, making this investment one of the largest private capital rounds in history. For comparison: the entire annual budgets of many countries are smaller than this sum. xAI, meanwhile, raised 20 billion dollars in Series E in January alone, showing that the pace of capital acquisition is accelerating, not slowing down.

This phenomenon creates a paradox for venture capitalists. On one hand, investments in these mega-rounds offer access to companies that could potentially change the world. On the other hand, returns from such capital are mathematically limited. If a company is already valued at hundreds of billions, the growth potential is much smaller than in earlier-stage investments. Venture capital traditionally sought 10x, 100x, sometimes even 1000x returns. At valuations approaching a trillion dollars, even 2x or 3x returns are already a significant achievement.

Returns: will the hype translate into real profits?

Here is the key question that few people ask out loud: do these gigantic AI investments actually generate profits, or are investors participating in the biggest speculative bubble since the dot-com era? The answer is complicated because most of these investments are too recent to assess real returns.

However, there are certain signals. Companies like OpenAI show the ability to generate revenue — the ChatGPT Plus subscription model, API access, corporate deals — these are not abstract promises but actual revenue streams. Anthropic, though less transparent in its finances, also shows growing corporate interest in Claude. But revenue does not equal profitability, and profitability is what ultimately matters for venture returns.

The problem is that the operating costs of these companies are astronomical. Training LLM models requires enormous computing power, and the costs of hardware, energy, and talent are high. OpenAI, for example, has many indications that it operates at an operating loss despite significant revenues. This means that even if a company raises hundreds of billions of dollars, it may never become profitable in the traditional sense — at least not in the near future.

Venture capitalists, however, are betting that these companies will eventually reach a scale that allows them significant profit margins. This is a gamble — but a gamble based on the logic that if any of these companies becomes the dominant AI platform, the returns will be so enormous as to justify every current loss. History shows that Google and Amazon operated at losses or minimal profits for years before becoming cash-generating machines. The question is: will OpenAI, Anthropic, and xAI be the next Amazons, or the next WeWork?

Why venture capital cannot resist AI

To understand why 41% of all venture money goes to AI, you need to understand investor psychology. In venture capital, there is a simple rule: if everyone is investing in something, you must too, or you risk being left out of the game. This is herd dynamics on steroids.

Additionally, AI has something few technologies have had — a clear potential to transform every industry. While other technologies (blockchain, IoT, quantum computing) remain niche or speculative, AI is already here, already working, already generating results. ChatGPT reached a million users faster than any other product in history. This is not hype without basis — this is hype based on real achievements.

Venture capitalists also know that if they miss an investment in the next "Unicorn" — a company that will change the world — they can lose their reputation. Better to invest in ten AI startups, nine of which will fail, if one becomes the next Google, than to sit on the sidelines and watch others collect trillions. This is the logic that drives capital concentration.

There is also a practical explanation: venture capital funds manage trillions of dollars. To generate the returns required by LPs (limited partners), they must make large investments. A startup that raises 50 million dollars may be interesting to a small fund, but for big venture players like Sequoia or Andreessen Horowitz, it's too small. Mega-rounds in AI are perfect for these large funds — they can invest a billion dollars and tell their LPs they have a position in the "next Google".

The remaining 59% of venture capital: an invisible catastrophe

If 41% of venture goes to AI, what happens to the remaining 59%? The answer is less dramatic, but equally important. The remaining funds go to traditional sectors: SaaS, fintech, healthtech, deeptech. But these are investments that receive far less media attention and far less investor optimism.

This creates a problem for the venture ecosystem. Entrepreneurs not working on AI face increasing difficulties in raising capital. Venture funds that traditionally invested in various sectors must now decide: follow the crowd to AI, or remain loyal to their historical areas of interest? Many chose the first option. Funds that once had diversified portfolios now have 50%, 60%, even 70% of their resources directed at AI.

This has long-term consequences. Innovation doesn't happen in just one sector. The best solutions to health, energy, or transportation problems may come from startups not working on large language models. But if these entrepreneurs cannot raise capital, they will not be able to develop their ideas. Venture capital, instead of financing innovation, is financing capital concentration in one sector.

Is this a bubble or a transformation?

The question everyone asks themselves privately is: is AI a real technological transformation or just a speculative bubble? History suggests it might be both at the same time.

Every major technology goes through a cycle: innovation, hype, bubble, crash, and then long-term growth. The internet went through this. Mobility went through this. Venture capital is always engaged in the hype and bubble phase — it's the nature of the business. Investors know there will be losses, but they count on profits from a few big winners compensating for those losses.

AI is different, however, because the technology is already here and already working. ChatGPT is not a promise — it is reality. Models like GPT-4 demonstrate capabilities that were impossible a few years ago. This is not hype without basis. But that doesn't mean prices can't be inflated. Reality and valuation are two different things.

My assessment is this: AI is a real technological transformation, but valuations are speculative. There will be a correction — maybe not tomorrow, but there will be. Some of these mega-startups will not survive. But a few will thrive and deliver returns that justify the entire investment. The problem is that venture capitalists don't know which ones will be winners, so they invest in all of them.

Implications for the technology ecosystem

When 41% of venture capital goes to one sector, it has a domino effect on the entire ecosystem. First, talent flows to AI. The best engineers, scientists, and managers are attracted to AI startups, where salaries and potential are greatest. This means other sectors struggle with recruitment.

Second effect: rental prices. In cities like San Francisco or Mountain View, where AI startups concentrate, rental prices have risen. Small companies that cannot compete with the salaries offered by OpenAI or Anthropic struggle to attract talent. This creates a gap between AI giants and the rest of the ecosystem.

Third effect: consolidation. When venture capital is concentrated, it favors consolidation. Large companies can afford R&D investments that small companies cannot. Large companies can also afford acquisitions. Result: the ecosystem becomes less competitive, more dominated by a few players.

For the long-term health of the technology industry, this concentration is dangerous. Innovation comes from competition, and competition requires a diversified ecosystem with many players. If venture capital continues to concentrate on AI, we risk creating a world where a few megafirms dominate and the rest of the ecosystem stagnates.

What this means for the future of venture capital

If current trends continue, we can expect the share of AI in venture capital to grow, not shrink. However, there is a saturation point. At some point, every investor who wants to invest in AI already does. When that happens, the dynamics may change.

Moreover, returns from AI investments may prove disappointing. If OpenAI never achieves profitability, if Anthropic struggles with competition, if xAI turns out to be another Theranos, venture capitalists may withdraw. History shows this industry changes its mind quickly when the numbers don't add up.

A realistic scenario is that in a few years we will look at this moment — when 41% of venture capital went to AI — as the peak of the bubble. But even if that happens, some of these investments will prove to be genius. Venture capital has always been a numbers game — if 90% of investments fail but 10% deliver 1000x returns, overall it was a good strategy. The question is: will AI be that 10%?