2025, the Year Artificial Intelligence Conquered Venture Capital

In 2025, something occurred that, until a few years ago, seemed difficult even to imagine: artificial intelligence (AI) absorbed about 50% of all global venture capital, raising 211 billion dollars, nearly double the 114 billion of 2024. This is not just a quantitative increase, but a structural shift in how venture capital perceives innovation, value, and growth.

The HumanX + Crunchbase 2025 AI Funding Report accurately captures this historical transition, demonstrating how AI is no longer a “technological gamble,” but the foundational infrastructure upon which entire industrial sectors are being redefined.

An Industry That Doubles in Just One Year

From 2016 to 2022, global investments in AI have progressively increased, with an initial acceleration in the post-pandemic period. However, it is between 2024 and 2025 that the real breakthrough occurs: +85% year-over-year, a dynamic rarely observed on a global scale.

The most significant data point is not just the total amount, but rather the structure of the rounds:

• $163 billion have flowed into rounds of $100 million or more
• 233 companies have closed megadeals
• These rounds represent 77% of all capital invested in AI in 2025, compared to 67% in 2024

Venture capital, in other words, is increasingly focusing on a few winners perceived as systemic.

Foundation Model: Dominant, But Not Alone

The foundation models remain the symbolic heart of the AI ecosystem. In 2025, they raised 87 billion dollars, with a growth of nearly 180% compared to the previous year.

OpenAI and Anthropic alone have attracted 58.5 billion, consolidating valuations that place them among the largest private companies in the world:

• OpenAI: estimated valuation around 500 billion dollars
• Anthropic: approximately 183 billion dollars

However, perhaps the most interesting data point is that 59% of AI investments did not go to foundation models, but to everything that makes AI usable, scalable, and monetizable.

Where Capital Really Goes: Infrastructure, Applications, and Deep Tech

Analyzing rounds above 100 million dollars reveals a much more complex distribution:

• 19% towards AI infrastructure (cloud, data labeling, platforms)
• 15% towards vertical AI software, with a particular focus on healthcare and security
• 11% towards deep tech, particularly robotics and defense

This shift indicates a market maturation: the focus is moving from pure computational power to the creation of measurable value.

Not surprisingly, several industry leaders emphasize that the issue is not ambition, but the foundations. According to research cited in the report, 95% of AI pilot projects do not produce measurable ROI, often due to infrastructural or organizational shortcomings. Companies that manage to bridge this gap are currently achieving average returns between 15% and 20%, with rapid improvement margins.

Geography of Power: The United States (and the Bay Area) Dominate

The geographical concentration of investments is impressive:

• 79% of all AI capital in 2025 went to U.S. companies
• $166 billion invested in the USA

Within the United States, the San Francisco Bay Area remains the absolute epicenter:

• 60% of global AI funding (approximately 126 billion dollars)
• 81% of all regional startup capital invested in AI
• 92 companies with rounds over 100 million

Yet, the Bay Area accounts for only 22% of the total number of deals, indicating that the global ecosystem is vast, but capital concentrates where iteration speed, talent, and capital collide more rapidly.

Women and AI: A Statistic That Requires Careful Consideration

One of the most surprising data points of 2025 concerns the presence of female co-founders in AI-funded companies in North America and Europe:

• 47% of AI capital has flowed into companies with at least one female founder
• Total of $84.7 billion

However, the report calls for a critical reading: the effect is heavily influenced by the mega-rounds of foundation models. Looking at the number of rounds, the percentage stabilizes around 20%, consistent with previous years.

The signal is positive, but it highlights how structural parity is still a long way off.

HumanX: Where Capital Meets Narrative

The report is not just a macro analysis, but also a snapshot of the ecosystem surrounding HumanX, the global summit dedicated to enterprise AI.

The over 130 companies taking the stage have raised more than 72 billion dollars since 2018. Among them are names like Databricks, Cerebras Systems, Synthesia, Runway, Cohere, and many others, active in sectors ranging from cloud to semiconductors, from generative video to automated coding.

HumanX positions itself as a “collision” space: data, founders, and investors don’t just discuss trends, they put them to the test in the field.

Looking Ahead to 2026: IPOs, M&A, and New Liquidity

Thanks to Crunchbase’s predictive intelligence, the report also attempts to look ahead:

• Out of approximately 6,600 AI companies funded since 2023, over 2,300 are considered likely acquisition candidates
• 443 companies show a high likelihood of IPO
• Among the companies present at HumanX, 27 could go public and 30 could be acquired
• Over half are expected to raise new rounds in the short term

After years of slowing exits, 2026 could mark a tangible reopening of the market.

Conclusion: not a bubble, but a historical reallocation

The message emerging from the HumanX + Crunchbase 2025 AI Funding Report is clear: AI is not merely experiencing a phase of hype, but a structural reallocation of capital.

Venture capital is betting on companies that:

• solve complex problems,
• generate measurable value,
• build lasting infrastructures.

In this sense, 2025 is not just the year when AI “captured” venture capital. It is the year when venture capital acknowledged that the future of innovation almost entirely stems from there.

Megadeals and Capital Concentration: The New Venture Paradigm

One of the key elements that distinguishes 2025 from previous years is the intense concentration of capital. Venture capital is not just investing more in AI: it is investing more selectively.

Megadeals (rounds exceeding 100 million dollars) have become the dominant tool for financing AI innovation. This results in two structural effects:

1. Reduction of perceived risk: large funds prefer to double down on already validated companies rather than fragment capital across dozens of early-stage bets.
2. Building systemic champions: many AI companies are not conceived as mere startups, but as future infrastructural layers of the digital economy.

This model is more reminiscent of the industrialization of the 20th century than the “spray and pray” venture approach of the 2010s.

AI as Economic Infrastructure, Not as a Feature

A key point of the report is the shift in narrative: AI is no longer treated as an additional feature, but as primary economic infrastructure.

Companies that attract significant capital in 2025 share certain characteristics:

• direct or privileged control of proprietary data;
• strong integration with core processes (supply chain, compliance, healthcare, security);
• business models oriented towards enterprise recurring revenue;
• ability to demonstrate measurable operational improvements.

In this context, AI becomes comparable to electricity or the Internet: invisible to the end user, yet essential for competitiveness.

The Topic of ROI: From Myth to Metric

The report addresses one of the most sensitive issues of AI adoption: the return on investment.

According to the cited data, 95% of AI pilot projects fail to deliver a measurable ROI. Not because the technology doesn’t work, but because there is a lack of:

• integration with legacy systems;
• data governance;
• internal training;
• redefinition of decision-making processes.

Companies that surpass this initial phase, however, enter a virtuous cycle. The current average ROI, estimated between 15% and 20%, is set to grow rapidly thanks to:

• reduction of marginal inference costs;
• model improvement;
• standardization of AI pipelines.

For venture capital, this means one thing: less hype, more execution.

Invisible Infrastructures: The Real Battleground

If foundation models represent the tip of the iceberg, AI infrastructures are the submerged mass.

In 2025, an increasing share of capital flowed into:

• cloud specialized for AI workloads;
• high-density energy data center;
• data labeling and synthetic data companies;
• chips and semiconductors optimized for training and inference.

These investments are less visible in the media, but often more defensible in the long term. They build technological moats that are difficult to replicate and bind customers through high switching costs.

Vertical Applications: When AI Becomes Business

Another sign of maturity is the growth of vertical applications.

Healthcare, cybersecurity, legaltech, defense, and finance are among the sectors that attract the most capital because they combine:

• high regulatory complexity;
• need for automation;
• availability of enterprise budget.

Here, AI is not experimentation, but a direct competitive advantage. Companies that manage to deeply integrate it into their workflows quickly become difficult to replace.

United States vs the Rest of the World: A Widening Gap

The U.S. dominance is not only quantitative but also qualitative.

The United States focuses on:

• the largest VC funds;
• the most advanced universities and research centers;
• big tech companies capable of acquiring or funding AI startups.

The result is a flywheel effect: more capital generates more talent, which creates more companies, attracting even more capital.

The rest of the world remains active and innovative, but faces increasing difficulties in competing in late-stage rounds.

Bay Area: Global AI Laboratory

The Bay Area emerges as a true global laboratory.

Here they focus on:

• the most advanced foundation models;
• the most sophisticated cloud infrastructures;
• an ecosystem of serial founders and specialized investors.

The most emblematic data is that 81% of regional startup capital has gone to AI. This means that, in fact, the Bay Area is betting almost exclusively on this technology as a driver of future growth.

The Role of HumanX: Validation Platform

HumanX is not just a conference, but a market validation platform.

The companies taking the stage are not concepts, but realities that:

• generate revenue;
• have enterprise clients;
• attract significant rounds.

This makes HumanX a prime observatory for understanding which business models are truly working.

Predictive Intelligence: Venture Capital Looks Ahead

The integration of Crunchbase’s predictive intelligence introduces a new element: the systematic forecasting of financial events.

Through the analysis of billions of signals, Crunchbase is able to estimate:

• probability of new rounds;
• likelihood of acquisition;
• IPO potential.

The fact that thousands of predictions have already been confirmed suggests a paradigm shift: venture capital no longer merely reacts, but seeks to anticipate.

IPO and M&A: Towards a New Liquidity Window

After years of contraction in exits, 2026 could represent a turning point.

The more mature AI companies demonstrate:

• stronger growth metrics;
• clear monetization paths;
• growing interest from corporates and public markets.

This could unlock new liquidity, reactivating the entire venture cycle.

Extended Conclusion: AI as the Architecture of the Future

2025 marks a dividing line.

Artificial intelligence is no longer a promise, but a backbone of the global economy. Venture capital has understood this and has reallocated resources accordingly.

Not all companies will succeed. The selection will be tough. But one thing is clear: the future of innovation, productivity, and industrial competitiveness will largely stem from here.

The year 2025 was not only when AI captured venture capital. It was the year when capital agreed to transform itself.