MMOOZZGG.com

AI investment

AI spending forecast for 2026 and 2027

6 min read4 sources

The Gartner AI spending forecast: $2.59 trillion in 2026 and $3.49 trillion in 2027. More than half of this money goes to infrastructure, not to software that businesses use. Below is a review of the forecast, its risks and conclusions for companies in Russia.

On this page · 5 sections

The Gartner forecast

Gartner estimates world AI spending in 2026 at $2.59 trillion. That is 47% more than in 2025, when spending was $1.76 trillion.

The forecast for 2027 is $3.49 trillion. If it comes true, AI spending will almost double in two years.

The 2027 figure, $3.49 trillion, is further from today and rests less on signed contracts. It is wise to read it as a direction, not as an exact sum.

The forecast was published in May 2026. Gartner updates such estimates, so we give the publication date and check for a new version before each update of the guide.

It is useful to know what Gartner counts. It is not investment in startups but all AI spending: equipment, services and software bought by companies, governments and people.

So you cannot compare the Gartner total with venture statistics. Venture data shows how much investors put into companies; Gartner shows how much AI buyers spent.

For a company, the forecast is useful as a background for the budget. World AI spending is growing, and suppliers will offer more solutions. Your task is not to spend more, but to spend on processes with a measurable return.

Where the money goes: infrastructure, services, software

According to Gartner, AI infrastructure will receive $1.43 trillion of the $2.59 trillion in 2026. More than half of all world AI spending goes to data centers, servers and chips.

The rest is split between services, software and devices. In the guide we give only the part of the Gartner breakdown we checked against the primary source: infrastructure.

Why infrastructure takes so much. Models need computing power, and suppliers build capacity before they sell AI services. Companies that buy AI solutions pay for this building through the price of subscriptions and plans.

For business, this leads to a simple split. Giants and governments build infrastructure. Mid-size companies get their return at the application level: where AI is built into sales, request handling, support and accounting.

Applied solutions get the smaller part of world money, but they are what gives results in a specific company: less manual work, faster answers to clients, fewer lost requests.

Look at your IT budget the way Gartner looks at the world: what share goes to infrastructure and licenses, and what share to solutions that change processes. If almost everything goes to the first part, it is too early to expect a return from AI.

Who invests how much in computing is covered in the article on AI infrastructure.

Hyperscaler capital spending

Hyperscalers, the largest data center owners, plan the following capital spending in 2026: Amazon about $200 billion, Microsoft about $190 billion, Alphabet $180–190 billion, Meta $125–145 billion.

Together that is $700–725 billion against $410 billion in 2025, growth of about 75%, according to a Forbes estimate.

According to an estimate in an industry review, hyperscaler capital spending grew by about 80% in a year, and more than $1 trillion is expected next year.

The two estimates give similar growth: about 75% and about 80% a year. Spending on capacity grows faster than total AI spending in the Gartner forecast, where growth is 47%.

This money is already tied to construction and equipment contracts, so the capital spending of the giants changes more slowly than startup valuations. It is a steadier guide than news about rounds.

A review of each company and each quarter is in the article on the capital spending of the giants.

What this means for you: a few American companies build the capacity, and Russian companies have limited access to their clouds. So for business in Russia, models that run on servers in Russia matter, and so do solutions that can move to such models.

Risks of the forecast

The Gartner forecast rests on the current plans of buyers. If investors and corporations cut their investment, AI spending will grow less than expected.

A report by Roscongress, a Russian organizer of business forums, put the chance of a crash of the AI company market in 2026 above 50%. This is the estimate of one report; read it as a risk signal, not as a forecast.

The same report gives concentration figures: since the start of 2024 the S&P 500 index grew by about $7.5 trillion, and 17 AI-related companies gave $4.9 trillion of that growth.

The report also describes circular financing: chip makers invest in clients who then buy their equipment. Such a scheme supports demand while the money goes around, and weakens it if one participant cuts investment.

For a company that buys AI, market risk looks different than for an investor. If some suppliers do not survive a correction, you lose not shares, but a working service.

To assess a supplier’s stability, ask three questions: who its investors are, which model it runs on and what happens to your data if it closes.

The protection is simple: the source code and settings are yours, the data is yours, and the solution can move to another model supplier. Signs of overheating and ways to protect yourself are covered on the page on the risks of AI investment.

What this means for companies in Russia

The world’s $2.59 trillion of spending creates the models and capacity that many AI solutions run on, including those sold in Russia. The prices and capabilities of these solutions depend on how world spending develops.

Three conclusions for the budget:

1. Do not plan AI from news about trillions. Count from processes: what lost requests, manual data entry and slow answers to clients cost.

2. Pay for a result in a process, not for access to technology. A contract with a metric protects you from paying for someone else’s infrastructure and getting no return.

3. Keep your independence: a solution on your own servers or one you can move. If the growth in the giants’ capital spending, $700–725 billion in 2026, turns into cuts, you will still have a working system.

4. Review the plan once a quarter. World forecasts change, and an AI budget approved for a year without review falls behind the market.

All forecast figures and related market data, with sources and dates, are on the statistics page.

Sources and fact-checking

Checked by the editors on 14 October 2026.

  1. Gartner, press release, 19 May 2026 (gcom.pdo.aws.gartner.com)
  2. Forbes, Jason Kirsch, 7 October 2026 (forbes.com)
  3. Blog of a major Russian cloud provider, webinar A look at the AI market, 24 July 2026 (yandex.cloud)
  4. Mail.ru Finance, citing RIA Novosti and a Roscongress report, 19 February 2026 (finance.mail.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