Расходы OpenAI на ИИ-инфраструктуру выросли до $750 млрд — это ВВП Швеции

OpenAI will direct around $750 billion toward computing infrastructure by 2030 — a sum comparable to Sweden's annual GDP. This estimate of the company's aggregate commitments was published by TechCrunch on July 22, 2026.

Where the $750 billion comes from

The $750 billion figure is the sum of OpenAI's previously announced computing capacity commitments, combined into one total. According to TechCrunch's calculations, the company's contracts for chips, data centers, and cloud resources together add up to roughly $750 billion by 2030.

Over the past year and a half, OpenAI has built an entire network of hardware partnerships: the large-scale Stargate infrastructure project, agreements with chip manufacturers and data center operators. Each such deal is not a one-time purchase but a multi-year capacity reservation, and it is their combined total that produces the resulting hundreds of billions.

• $750 billion — total infrastructure spending on the horizon through 2030
• The scale is comparable to Sweden's annual GDP
• The estimate was published by TechCrunch on July 22, 2026
• The bulk of the spending is computing capacity: chips, data centers, cloud contracts

Why does it cost so much?

Training and launching frontier models runs up against a compute shortage: each new generation of AI requires multiples more GPUs, electricity, and data center space. To secure this capacity years in advance, OpenAI signs multi-year contracts worth tens and hundreds of billions of dollars, effectively reserving future supplies of chips and server racks ahead of time.

Spending on this scale is reshaping the economics of the entire industry. $750 billion is no longer the budget of a single product — it is a sum on the scale of a national economy, spread over several years and tied to chip suppliers, data center operators, and energy providers.

What's behind the compute race

The key scarce resource of the AI industry has become not ideas, but access to computing — GPUs, electricity, and data center space. Companies that reserve capacity in advance gain an advantage in training the next generations of models.

That's why these commitments are measured in sums on the scale of national budgets. The $750 billion, spread out through 2030, ties OpenAI to chip suppliers, data center builders, and energy companies long before this capacity is actually needed.

"OpenAI will spend on infrastructure a sum equivalent to Sweden's GDP,"

TechCrunch's article from July 22, 2026 states.

What this means

The race among AI labs has entered a stage where the price of entry is measured in hundreds of billions of dollars, and the key resource is access to computing rather than the algorithms themselves. OpenAI's $750 billion in commitments through 2030 shows that maintaining leadership in AI now requires capital on the scale of entire nations.

Frequently Asked Questions

How much will OpenAI spend on infrastructure?

According to TechCrunch's estimate, around $750 billion on the horizon through 2030 — a sum comparable to Sweden's annual GDP.

What will this money be spent on?

The bulk of the spending is computing capacity: purchasing chips, building and leasing data centers, and long-term cloud contracts for training and launching AI models.