AI Spending Statistics 2026: $2.67 Trillion & Big Tech Capex

Last Updated: October 20, 2026

Worldwide AI spending will reach $2.67 trillion in 2026, up 49.5% from $1.79 trillion in 2025, according to Gartner’s September 2026 forecast. Gartner expects it to pass $3.6 trillion in 2027. The four biggest tech companies, Amazon, Alphabet, Microsoft and Meta, have guided to a combined $720 billion to $745 billion in capital spending for 2026, most of it for AI data centers. And Nvidia, which sells most of the chips inside those data centers, booked $89.0 billion in data center revenue in a single quarter.

So how much are companies spending on AI in 2026? At the company level, the numbers are far smaller and far more uneven. The median US business on Ramp’s platform spends about $12.50 per employee per month on AI tools, while the top 10% spend $675.60. Companies plan to spend an average of 1.7% of revenue on AI in 2026, more than double the 2025 level, according to BCG.

The most important finding in the 2026 spending data: money is moving much faster than returns. 86% of enterprises expect their AI budgets to rise this year, according to NVIDIA, yet 56% of CEOs say they have seen no significant financial benefit from AI so far, according to PwC.

The money is flowing at a scale never seen before. The question is whether it is flowing to the right places.

Everything about AI spending on AI Business Weekly

Global AI spending figures in 2026 come from several research firms using different methods, so the totals vary a lot. Know which number measures what before using one in a business case.

The main measurements:

These numbers do not contradict each other. Gartner counts almost everything AI touches, including the AI built into new phones, PCs and servers. IDC counts only AI infrastructure hardware. Capex counts what the biggest cloud companies spend on buildings, servers and chips. Stanford counts money invested into AI companies.

The growth trajectory:

Gartner has raised its forecast twice in 2026. In January it called for about $2.5 trillion. In May it raised that to $2.59 trillion (+47%), and in September to $2.67 trillion (+49.5%). Its 2027 forecast is $3.64 trillion. Gartner analyst John-David Lovelock said AI infrastructure demand “remains strong and inelastic to pressures from memory-related pricing increases.”

Gartner’s worldwide AI spending forecast (September 2026):

For how this spending translates into actual company use, see our AI adoption statistics.

AI capex, the money big tech companies spend on data centers, servers and chips, is the largest and fastest-growing part of AI spending in 2026. Each of the four biggest companies raised or held its plans through the year.

2026 capex guidance by company:

Sources: Amazon via Fortune, Alphabet via W.Media, Microsoft via Reuters/The Star, Meta via Blockspace, combined tracker via Yield Theory. Each company defines capex slightly differently, so treat the combined figure as a range.

Why the numbers keep going up:

AI infrastructure spending in Q1 2026:

AI infrastructure spending reached $89.7 billion in Q1 2026 alone, up 33.1% year over year, according to IDC’s Worldwide Quarterly AI Infrastructure Tracker. Based on that result, IDC raised its full-year 2026 forecast to $497 billion, up about 56% from $318 billion in 2025. Servers made up 97.6% of the Q1 total.

The biggest shift in Q1 2026: Arm-based accelerated servers ($53.0 billion) overtook x86 ($34.6 billion) for the first time, according to IDC. That matters for which chip and server vendors win the next wave of contracts.

Where the money lands: Nvidia

Nvidia reported $96.2 billion in revenue for the quarter ending July 26, 2026, up 106% from a year earlier, with $89.0 billion from data centers, up 117%, according to Nvidia’s Q2 fiscal 2027 results. It expects about $108 billion in revenue next quarter. A large share of Big Tech’s AI capex ends up as Nvidia revenue. See our Nvidia AI statistics.

The single largest AI infrastructure commitment outside the hyperscalers comes from a partnership.

Stargate is a joint venture between OpenAI, SoftBank and Oracle, announced on January 21, 2025, to invest up to $500 billion in US AI data centers over four years, starting with $100 billion. SoftBank leads the financing, Oracle builds and runs much of the infrastructure, and OpenAI is the main customer for the computing power.

Stargate by the numbers:

Source: WebProNews, April 2026, summarizing OpenAI announcements and reporting from Bloomberg, The Wall Street Journal and Reuters.

Stargate’s path has not been smooth. In mid-2025, reports described disputes among the partners over control, and no major deals had been signed six months after the launch. The project then sped up, with new sites announced and construction under way. Most of the figures above are plans and commitments, not finished capacity.

The supply constraint money cannot solve:

The biggest limit on AI infrastructure in 2026 is not money. It is power, chips, memory and construction time. Amazon says it still cannot meet AI demand, and memory price increases pushed its 2026 spending up by about $20 billion. Even committed capital can only be deployed as fast as data centers can be built and powered.

For OpenAI’s side of the story, see our OpenAI statistics and OpenAI IPO guide.

Below the hyperscaler level, company AI spending is rising fast, but with a big gap between leaders and everyone else.

The core enterprise spending numbers:

How much of the IT budget goes to AI?

28% of organizations now spend more than 10% of their tech budget on AI, according to McKinsey’s 2026 survey. At the industry level, Gartner’s July 2026 IT forecast shows data center systems, the category most tied to AI, growing 62.5% to $822 billion in 2026, while total IT spending grows 14.2% to $6.37 trillion. AI is taking a fast-growing share of every IT dollar.

Where enterprise money goes by provider:

Anthropic held 40% of enterprise spending on AI models in 2025, OpenAI 27% and Google 21%, according to Menlo Ventures. By September 2026, 43.8% of US businesses paid for Anthropic products and 40.6% for OpenAI’s, according to the Ramp AI Index. See our AI market share guide for the full breakdown.

The cost pressure is starting to show:

About 20% of McKinsey’s 2026 respondents say AI operating costs, including token costs, have limited how much they use AI. For deeper data on how much companies use AI, see our AI adoption statistics.

The per-employee number is the most useful benchmark for finance leaders. It answers the question: “Are we spending about the right amount on AI compared with our peers?”

The best public data comes from Ramp, which tracks real card and bill payments from more than 70,000 US businesses.

Monthly AI spend per employee (Ramp, August 2026):

Source: Ramp’s August 2026 AI Index data, as reported by Accounts Recovery. Annual figures are simple monthly x 12 estimates.

The gap is enormous. The top 10% of spenders pay about 54 times the median per employee, and the top 1% pay more than 500 times. Most companies are still buying a few chat subscriptions. A small group is running AI agents and coding tools that consume large amounts of computing every day.

By company size:

Small businesses spend the most per employee, because a few AI subscriptions are a bigger share of a small headcount. Large companies spread enterprise deals across many more employees.

The shadow spend problem:

Official budgets miss a lot of AI spending. Employees often pay for their own AI tools, and teams can rack up usage-based costs quickly. Uber is the best-known example of 2026: COO Andrew Macdonald said in May that the company had already used up its 2026 AI budget, driven by rising token costs tied to Claude Code, and that the company was struggling to link those costs to more consumer features, according to Quartz. The lesson for every CFO: usage-based AI tools need usage tracking, not just an annual budget line.

Earlier versions of this page cited an average of $2,068 per employee from the Federal Reserve Bank of Atlanta. We could not find that figure in any Atlanta Fed publication, so we have replaced it with Ramp’s transaction data.

💡 Not sure which AI tool is actually right for your business?

Get the free guide, Which AI Tool Should You Actually Use? — a straightforward breakdown of the leading AI tools to help you pick the right one for your needs.

Subscribe to AI Business Weekly for the guide, plus daily coverage of AI trends, acquisitions, and product launches.

Get the Free Guide →

Industry predicts AI spending intensity better than company size does.

Median monthly AI spend per employee by industry (Ramp, August 2026):

Source: Ramp’s August 2026 AI Index data, via Accounts Recovery.

Tech and media companies spend about 8 times as much per employee as manufacturers. Finance and professional services sit in between.

The same order shows up in adoption. 39.7% of US information businesses and 33.9% of finance and insurance businesses use AI, compared with about 14% in retail, according to the US Census Bureau. In Canada, information and cultural businesses (42.3%) and finance and insurance (40.4%) lead, while agriculture (4.5%) and construction (9.2%) trail, according to Statistics Canada.

Where AI agents are spreading fastest:

Telecommunications had the highest share of companies deploying or assessing AI agents, at 48%, followed by retail and consumer packaged goods at 47%, according to NVIDIA’s State of AI 2026. Across all industries, 44% were deploying or assessing agents.

For industry deep dives, see our statistics on AI in finance, AI in healthcare, AI in manufacturing and AI in retail.

How AI budgets are split matters as much as how big they are.

At the market level, Gartner’s 2026 forecast shows where the $2.67 trillion goes:

Source: Gartner, September 2026. Shares calculated from Gartner’s figures.

What the split tells you:

Inside companies:

Public data on how individual companies split their AI budgets is thin, so be careful with neat percentages you see online. What the strongest research does show is where the payoff comes from. About three-quarters of McKinsey’s AI high performers have fundamentally redesigned their workflows around AI, compared with about a quarter of other companies. Spending on implementation and change, not just licenses, is what separates the 6% of companies getting real profit from AI from everyone else.

The governance line is growing:

With Article 50 of the EU AI Act enforceable from August 2, 2026, companies that operate in Europe must label AI-generated content and tell users when they are talking to AI. Fines can reach €15 million or 3% of global turnover, which is pushing more budget toward AI governance and compliance.

The most important finding in the 2026 AI spending data is not how much is being spent. It is how little of that spending is producing clear returns so far.

The ROI gap in numbers:

Sources: NVIDIA via IT Brief, PwC and Tech.co, McKinsey.

Why NVIDIA and PwC seem to disagree:

NVIDIA’s survey asked whether AI has increased revenue or cut costs in some or all parts of the business, a low bar that any single win can meet. PwC asked 4,454 CEOs whether they had seen significant financial benefit, a much higher bar. Both can be true: AI is producing many small wins, but few companies see it in the bottom line yet.

The returns when they do show up:

An IDC study for Microsoft of more than 4,000 business leaders found an average return of $3.70 for every $1 invested in generative AI, with top leaders reporting $10.30, according to Microsoft. That study dates from November 2024 and was sponsored by Microsoft, so treat it as an upper-end view.

At the other end, MIT’s NANDA initiative found 95% of enterprise organizations got zero return from their generative AI efforts, despite $35 billion to $40 billion in US investment, as reported by The Register.

The pattern behind both numbers:

Returns concentrate in a small group. McKinsey’s 6% of high performers are the ones that redesigned workflows, scaled AI across the company and track profit impact. Most companies are still buying tools and running pilots. Spending more does not close the gap on its own. Changing how work gets done does. For the full picture, see our AI ROI statistics and AI productivity statistics.

Small and medium businesses spend on AI very differently from large companies: mostly subscriptions, not infrastructure, and with tighter budgets.

The SMB spending picture:

What the gap means:

Small businesses that adopt AI spend heavily for their size, but many small firms have not started at all. Among those that use AI, most spending goes to subscriptions like ChatGPT, Claude, Gemini or AI features built into tools they already use.

Why the SMB return case is simpler:

The math is easier for a small business. A team that pays for a few AI subscriptions can compare that cost with hours saved each week. There is no long integration project to finish first. That is one reason small firms that adopt AI tend to commit to it.

For tools and budgets at the small business level, see our AI statistics for small business and best AI tools for small business.

AI spending splits into three big buckets, and they are not the same size.

Source: Gartner, September 2026. Shares calculated from Gartner’s figures.

Infrastructure is the biggest bucket by far, at more than half of the total. That is the data center buildout: servers, accelerators and the systems around them.

Software is growing fastest of the big three, up 60% in 2026, from $288 billion to $462 billion.

Earlier versions of this page said AI software was 52% of total AI spending. Gartner’s own figures put it at about 17%, so we have corrected it.

The agent category:

In its September 2026 forecast, Gartner moved AI agents and assistants into their own category, now including consumer agents. That category grows from $16.5 billion in 2025 to $29.2 billion in 2026 and $65.5 billion in 2027, more than doubling next year. Gartner also predicts more than 40% of agentic AI projects will be canceled by the end of 2027 because of rising costs, unclear business value or weak risk controls, according to Gartner. See our AI agents statistics.

The United States dominates AI investment by a wide margin.

AI investment by region, 2025 (Stanford AI Index 2026):

Source: Stanford HAI, 2026 AI Index Report.

US AI investment was more than 13 times Europe’s and 23 times China’s in 2025, by Stanford’s count. Global corporate AI investment hit a record $581.7 billion, up from $14.6 billion in 2013.

The same pattern holds for infrastructure. The US accounted for $67.9 billion, or 75.7%, of global AI infrastructure spending in Q1 2026, according to IDC.

Why China’s number looks so small:

Stanford’s figures track investment into AI companies, mostly from private investors and acquisitions. Much of China’s AI spending comes from large tech companies and the state, which these figures capture less fully. China’s AI models have nearly caught up anyway: the 2026 AI Index found the performance gap between top US and Chinese models had narrowed to 2.7%.

The sovereign AI trend:

A growing share of AI spending now comes from governments building AI computing capacity inside their own borders, driven by data control and national competitiveness. Stargate’s planned site in Argentina and the proposed 5 gigawatt campus in Abu Dhabi are examples of AI infrastructure deals that now involve national governments.

With spending forecast to pass $3.6 trillion in 2027, the question investors ask most is whether this is a bubble. The data points both ways.

Signs the spending is backed by demand:

Signs of strain:

The honest read: the infrastructure buildout is being paid for by real, fast-growing AI revenue at the model and cloud layer. Whether the companies buying AI get enough back to keep paying is the open question for 2027. See our AI bubble analysis.

The AI spending data in October 2026 has practical lessons beyond the headline numbers.

Lesson 1: The infrastructure buildout will shape AI prices for years.

$720 billion to $745 billion of Big Tech capex in 2026, plus Stargate, represents multi-year commitments. That capacity will set the cost and availability of AI services through the end of the decade. Capacity is still tight today, so expect AI prices to keep shifting as new data centers come online.

Lesson 2: Benchmark per employee, not by total budget.

The median US business spends about $12.50 per employee per month on AI. The top 10% spend $675.60. If you compare yourself with the median, you are comparing yourself with companies that are still experimenting. Decide which group you want to be in.

Lesson 3: Budget for the work around the tools.

The companies getting real returns redesign workflows, train their people and track results. Licenses alone do not produce the 6% high-performer outcome. Budget for implementation and change, not just subscriptions.

Lesson 4: Track usage, not just invoices.

Usage-based AI tools, especially coding agents and APIs, can blow through a budget fast, as Uber found. Set up spending visibility by team before scaling usage.

Lesson 5: Measure what you spend on.

56% of CEOs cannot point to significant financial benefit yet. The companies that can are the ones that defined what success looks like before they spent. See our how to implement AI in business guide.

Nvidia AI Statistics 2026
The chip company behind most AI infrastructure spending.

AI Adoption Statistics 2026
How many companies use AI, by size and industry.

Generative AI Market Statistics 2026
Market size projections that put the spending data in context.

AI ROI Statistics 2026
What companies are getting back from their AI spending.

AI Productivity Statistics 2026
What AI spending is producing in measurable output.

AI Agents Statistics 2026
The fastest-growing AI category and its deployment data.

Best AI Tools for Small Business 2026
How small businesses spend their AI budgets.

OpenAI Statistics 2026
Stargate and OpenAI’s revenue trajectory.

How much is being spent on AI globally in 2026?
Worldwide AI spending will reach $2.67 trillion in 2026, up 49.5% from $1.79 trillion in 2025, according to Gartner’s September 2026 forecast, and is expected to reach $3.64 trillion in 2027. More than half goes to AI infrastructure. Narrower measures are smaller: IDC forecasts $497 billion in AI infrastructure spending in 2026. Amazon, Alphabet, Microsoft and Meta together have guided to $720 billion to $745 billion in 2026 capital spending.

How much do companies spend on AI per employee?
The median US business spends about $12.50 per employee per month on AI, according to Ramp’s August 2026 data from more than 70,000 businesses. The top 10% of spenders pay $675.60 per employee per month and the top 1% pay $7,205.13. Small businesses spend a median $23.23 per employee per month, compared with $3.68 at large companies. BCG found companies plan to spend 1.7% of revenue on AI in 2026.

Which industry spends the most on AI in 2026?
Technology and media companies spend the most on AI per employee, at a median $80.15 per month, according to Ramp’s August 2026 data. Finance and insurance follow at $41.24, then professional, scientific and technical services at $33.33. Manufacturing spends $9.84. Adoption follows the same order: information and finance businesses use AI at roughly twice the national rate in both the US and Canada.

How much are Amazon, Google, Microsoft, and Meta spending on AI in 2026?
Amazon expects about $220 billion in 2026 capital spending, Alphabet $195 billion to $205 billion, Microsoft about $175 billion and Meta $130 billion to $145 billion, based on guidance given in July 2026. Together that is $720 billion to $745 billion, mostly for AI data centers, servers and chips. Amazon and Alphabet both raised their plans during 2026, citing AI demand and higher memory costs.

What is the Stargate AI program and how much does it cost?
Stargate is a joint venture between OpenAI, SoftBank and Oracle, announced in January 2025, to invest up to $500 billion in US AI data centers over four years. More than $400 billion had been committed by September 2025, and planned capacity reached nearly 7 gigawatts across sites in Texas, New Mexico, Ohio, Michigan, the Midwest and Argentina. The flagship campus in Abilene, Texas is in partial operation.

What is the ROI on AI investment in 2026?
Returns are real but concentrated. 56% of CEOs have seen no significant financial benefit from AI yet, according to PwC, and only about 6% of companies get 5% or more of their EBIT from AI, according to McKinsey. At the same time, 88% of enterprises say AI has increased revenue in some part of their business, according to NVIDIA. An IDC study for Microsoft found an average return of $3.70 per $1 invested in generative AI.

How much of the IT budget goes to AI in 2026?
28% of organizations spend more than 10% of their technology budget on AI, according to McKinsey’s 2026 survey. Across the market, Gartner forecasts data center systems, the category most tied to AI, will grow 62.5% to $822 billion in 2026, far faster than total IT spending, which grows 14.2% to $6.37 trillion. BCG found companies plan to spend 1.7% of revenue on AI in 2026.

How much does the average SMB spend on AI?
Small US businesses spend a median $23.23 per employee per month on AI, according to Ramp’s August 2026 data, more than six times the per-employee spend at large companies. 50.0% of small businesses on Ramp’s platform paid for AI tools in September 2026. Across all US businesses, including the smallest, only 19.8% use AI in their operations, according to the US Census Bureau.

The AI spending data in October 2026 tells two stories at once.

Story one is the infrastructure story. $2.67 trillion in worldwide AI spending in 2026. $720 billion to $745 billion in Big Tech capex. $89.0 billion in Nvidia data center revenue in a single quarter. $500 billion in Stargate commitments. This is the largest infrastructure buildout in the history of technology, and it is still speeding up, with every major forecast raised during 2026.

Story two is the company reality story. The median business spends about $12.50 per employee per month on AI. 56% of CEOs have not seen significant financial benefit yet. Only about 6% of companies get a meaningful share of profit from AI. A few companies, like Uber, are already hitting the limits of usage-based costs.

Both stories are true. The scale of the buildout shows the AI economy is real. The gap between spending and returns shows most companies are still early in learning how to use it.

The companies that look back on 2026 as a turning point will be the ones that closed that gap: spending with intent, redesigning how work gets done, and measuring what they get back.

By Sameer Khan

This article was AI-assisted, then reviewed and edited by Sameer Khan before publishing.

Sameer Khan is the founder of AI Business Weekly, a daily newsletter and research hub covering AI tools, funding and business adoption. He holds an MBA from the Ted Rogers School of Management and has spent nearly a decade in B2B sales across SaaS, research and AI. Connect on LinkedIn.