The Quarter AI Took 80% of Venture Capital── Breaking down mega-round concentration, strategic circular financing, and what regulators see

01AI now takes a larger share of VC than internet companies did at the dot-com peak

According to PitchBook, AI startups' share of global venture funding rose from 38% in 2024 to 52.5% in 2025 — the first year AI claimed a majority — and reached 80% in Q1 2026. At the peak of the dot-com boom in 2000, internet companies accounted for roughly 39% of venture dollars. The crypto peak in 2021 reached 22%. AI's current share is double the dot-com record.

Crunchbase reports that global venture investment in Q1 2026 totaled approximately $300 billion across about 6,000 funded companies, a year-over-year increase exceeding 150%. Of that total, $242 billion went to AI companies. But the distribution was sharply skewed: four companies — OpenAI ($122 billion), Anthropic ($30 billion), xAI ($20 billion), and Waymo ($16 billion) — collectively raised $188 billion, or roughly 65% of all global venture investment in the quarter.

02A handful of mega-rounds are moving the aggregate numbers

The 80% headline obscures a critical detail: capital is not flowing broadly across AI startups but concentrating in a few foundation-model companies. OpenAI's $122 billion round alone represented 40% of all global venture funding in Q1 2026. SoftBank led the round, which included commitments to the Stargate joint venture with Oracle for AI infrastructure buildout.

These rounds share a distinctive feature. A large portion of the capital is earmarked for compute — purchasing or reserving GPU capacity and cloud resources. When the primary use of venture funding is to buy computing power, the dynamics differ from traditional startup financing where capital funds product development, hiring, and market expansion. The capital intensity of training and running large models creates a self-reinforcing cycle: more capital enables more compute, which enables larger models, which attract more capital.

03The concentration of AI funding is narrowing the pool for everything else

If the total venture pool were growing proportionally, concentration might be less concerning. It is not. In Q1 2026, roughly $58 billion was left for non-AI startups — divided among thousands of companies. Stanford HAI's 2026 AI Index Report recorded that global corporate AI investment reached $581.7 billion in 2025, with private investment alone at $344.7 billion, up 127.5% year-over-year. The newly funded AI companies rose 71%, and billion-dollar rounds nearly doubled.

The same report noted that this growth came disproportionately from the United States, which accounted for $285.9 billion — 23 times China's $12.4 billion. The geographic concentration mirrors the corporate concentration: a small number of San Francisco-based companies, backed by a small number of large investors, in a single country. For sectors like biotech, clean energy, and fintech, the practical consequence is a shrinking share of a growing pie.

04Strategic investors who are also the primary suppliers create a circular flow

A structural feature of AI mega-rounds is the overlap between investor and supplier. Microsoft has invested a cumulative $13.8 billion in OpenAI and holds 27% of its for-profit entity. OpenAI has committed to purchasing up to $250 billion of Azure cloud capacity through 2032. Amazon has invested up to $8 billion in Anthropic. Google has invested approximately $2 billion in Anthropic. In each case, the AI company runs its workloads on the investor's cloud platform.

The result is a circular pattern: the cloud provider invests in the AI company; the AI company uses the investment to buy compute from the cloud provider; the cloud provider's revenue grows; the cloud provider's stock price rises; the cloud provider invests more. The FTC published a staff report in January 2025 examining this structure across three relationships — Microsoft-OpenAI, Amazon-Anthropic, and Google-Anthropic. The report raised concerns that contractual terms and technical barriers could make it difficult for AI developers to switch cloud providers, and that cloud providers may gain access to sensitive technical and business information through these arrangements.

05Not all of the announced amount is cash — compute credits change the calculus

When a mega-round is announced, the headline number may include cloud compute credits alongside cash. Amazon's investment in Anthropic includes AWS credits. Microsoft's investment in OpenAI has been structured with Azure usage rights. The proportion of cash to credits is not always disclosed.

From the startup's perspective, compute credits are as valuable as cash — training a frontier model requires massive GPU clusters regardless of the payment method. From the investor's perspective, the cost of providing cloud credits may be significantly below the retail price. A $10 billion investment that includes $4 billion in compute credits at a 60% internal margin represents a different economic commitment than $10 billion in cash.

The UK's Competition and Markets Authority reviewed the Google-Anthropic relationship in late 2024 and concluded it did not meet the threshold for a merger under UK competition law. However, the broader question of whether cloud providers become de facto preferred suppliers through their investment relationships remains an open area of regulatory inquiry across jurisdictions.

06Pharmaceutical AI investment is growing fast from a much smaller base

The AI venture capital story is dominated by foundation-model companies, but capital is reaching healthcare and pharmaceutical AI as well. According to BioPharma Dive, approximately 30% of healthcare venture funding in 2024 went to AI-enabled companies, with total AI healthcare investment reaching $5.6 billion — roughly triple the previous year. As of October 2025, more than 530 companies worldwide were focused on AI-powered drug discovery.

To put that in perspective: the entire AI healthcare venture total for 2024 was less than 5% of what OpenAI raised in a single quarter of 2026. The scale difference matters because it reflects different risk profiles. Pharmaceutical AI investments face a longer path to revenue — from molecule design through preclinical testing, clinical trials, regulatory approval, and commercialization. That timeline stretches 7 to 12 years, exceeding the typical VC fund cycle of 5 to 7 years.

As discussed in the third article of this series, the gap between investment and revenue in AI is a general phenomenon. In pharmaceutical AI, it is amplified by the regulatory timeline inherent to drug development. The number of AI-designed molecules entering clinical trials is rising, but no AI-designed drug has yet received regulatory approval. The investment thesis rests on the expectation that AI will compress early-stage discovery timelines by two to three years, but the full development cycle from candidate to market remains long.

07Three questions to ask when reading an AI funding announcement

When a large AI funding round is announced, three questions help distinguish substance from spectacle.

First, what is the composition of the capital? How much is cash, and how much is compute credits or cloud commitments? If the breakdown is not disclosed, that absence is itself informative. A round described as $10 billion that includes $4 billion in below-market-cost compute credits has different implications than $10 billion in cash.

Second, do the investors overlap with the customers? When a cloud provider both invests in and sells services to an AI company, the FTC report's concerns about lock-in and circular revenue become relevant. Reading the investor list alongside the supplier agreements reveals the extent of this overlap.

Third, how many stages separate the investment from revenue? The three-stage framework from the third article in this series — GPU maker revenue, cloud service revenue, end-user revenue — applies to infrastructure-focused rounds. For pharmaceutical AI, additional stages (clinical trials, regulatory approval, commercialization) extend the chain further. More stages mean more uncertainty.

This series tracks investment data as part of its educational mission. The goal is not to predict which investments will succeed, but to provide the structural understanding needed to read the numbers critically.