MMOOZZGG.com

AI investment

Corporate AI investment in startups: why giants buy stakes

5 min read5 sources

Corporate AI investment in startups reaches tens of billions of dollars in 2026: Amazon put $50 billion into OpenAI and Nvidia $30 billion. Below: why giants buy stakes instead of building their own, where the risk sits in this scheme and what a company in Russia can borrow from it.

On this page · 5 sections

Stakes instead of development

Amazon, Google and Nvidia have their own AI teams, and still they invest in other companies. In the OpenAI round of $122 billion, Amazon gave $50 billion, and Nvidia and SoftBank $30 billion each, Bloomberg Law reported.

Google did the same with Anthropic: in the May 2026 round of $65 billion, it put in several billion dollars under an earlier commitment.

The reason is simple. A stake in a company that already builds world-class models gives access to the technology at once. Your own development takes years and promises no result.

A stake gives a second thing: if the startup grows, the investor earns on the rise in valuation. After its round OpenAI was valued at $852 billion, and Anthropic at $965 billion.

Money in exchange for cloud and chips

Large investors in the OpenAI and Anthropic rounds also sell them what these startups spend money on. Amazon and Google sell cloud computing, and Nvidia sells chips.

The scheme looks like this: an investor puts in, say, $30 billion, the startup spends a good part of it on computing, and the money returns to the investor as revenue. The investor gets a stake, a client and sales growth at the same time.

For the startup, the plus is that it gets both money and secured access to capacity. The minus is dependence: its costs and plans are tied to one supplier.

Such deals read differently from ordinary venture capital. The size of the round shows not only faith in the company but also the volume of future orders the investor has secured.

For an AI buyer, this raises a question for your contractor: who are its investors, and is it obliged to buy computing from one of them? The answer decides whether the contractor can move your solution to another supplier if prices rise.

Circular financing and its risks

This scheme has a name: circular financing. The Roscongress report describes it this way: chip makers invest in clients who buy their equipment.

The risk is that demand in such a chain is partly created by the seller’s own money. If the rounds stop, clients cut purchases, and the supplier’s revenue falls together with the startup’s valuation.

Roscongress also links this to a general market risk: the report puts the chance of a crash of the AI company market in 2026 at above 50%. Since early 2024 the S&P 500 grew by about $7.5 trillion, and $4.9 trillion of that came from 17 AI-related companies.

We do not forecast share prices. For a company that buys AI, the lesson is practical: check that your supplier does not depend on one investor and one source of money.

The page on the risks of AI investment covers the signs of overheating and how to protect your own investment.

Buying teams and technology

A stake is not the only path. In the third quarter of 2026, giants bought whole companies, per Crunchbase:

1. Nvidia offered to buy Hugging Face for $12.9 billion.

2. AMD announced its plan to buy World Labs for $8.2 billion.

3. Stripe bought OpenRouter for $7.5 billion.

Each purchase solves its own task. Hugging Face runs the platform where developers publish models, World Labs builds models of 3D worlds, and OpenRouter gives single access to many models.

The price of a purchase shows how much the buyer values the time it saves. The $12.9 billion offer for Hugging Face is the price of a ready platform with a developer community that nobody can build quickly from zero.

The buyer pays for a team and a product that already work, and saves years of development. A separate page explains how buying a stake in an AI company works for a smaller business.

What applies in Russia

In Russia, too, AI money comes mostly from large companies, not from venture funds. Per a 2026 industry study, large Russian IT corporations publicly put $2.7 billion into AI in 2025, while venture investment in AI companies was about $51 million.

The ratio of corporate to venture money in this study is about 52:1. Experts say the method needs checking, so read the figure as a direction, not as an exact measurement.

Three methods of the world’s giants suit a company in Russia:

1. Investment in exchange for orders. If a startup builds a solution for your industry, first become its client with a pilot and a metric, and talk about a stake after the result.

2. Buying a team instead of building. A ready product with working clients is often faster than your own development from zero.

3. A corporate fund. When deals multiply, they go into a separate fund with rules for selection and checks; see the page on the corporate venture fund.

It is too early to start with buying a stake while it is unclear where AI makes money for your company. First find processes with a measurable effect, then decide whether to build, buy or invest.

Sources and fact-checking

Checked by the editors on 14 October 2026.

  1. Bloomberg Law, 1 April 2026 (news.bloomberglaw.com)
  2. Bloomberg Law, 28 May 2026 (news.bloomberglaw.com)
  3. Crunchbase News, 5 October 2026 (news.crunchbase.com)
  4. Finance Mail, citing RIA Novosti and a Roscongress report, 19 February 2026 (finance.mail.ru)
  5. ComNews, citing a Kama Flow study, 6 October 2026 (comnews.ru)

Found a mistake? Tell us

Publications

Mike Mozg

Agentic AI use cases at AT&T: nearly 500 processes now run on agents

AT&T no longer uses AI only to help staff work faster: it is rebuilding whole business processes around agents. SiliconANGLE reported on 14 November 2026 that nearly 500 agentic AI use cases at AT&T are already in production. Here is how the company rebuilds them, where a human still checks the agent, and what other large companies can take from it.

News4 min
Contact us