The race is heating up. Elon Musk’s SpaceX, which builds AI models as well as space rockets, announced last week that it’s seeking a $1.77 trillion (£1.31 trillion) valuation on the US stock market. Meanwhile, Anthropic, the startup behind the Claude chatbot, said it had filed for an initial public offering. OpenAI, the developer of ChatGPT, is expected to follow suit.
This latest peak in the AI market comes amid a multi-trillion-dollar spending spree on related infrastructure, like data centers. At the same time, companies are trying to use the technology in ways that make their investments worthwhile. Here’s a look at where the AI boom stands and six key charts that show how we got here.
1. AI has driven stocks to new heights
The S&P 500, which tracks the 500 largest US companies, has surged nearly 80% over the past five years. That jump has been fueled by big tech stocks with a stake in the AI boom—the “magnificent seven”: Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla.
Investor focus on technology is unprecedented, says Jim Bianco of US firm Bianco Research. His research found that 41 AI-related stocks now account for nearly half of the S&P 500’s total market value.
Neil Wilson, an analyst at investment platform Saxo UK, warns that the risk of a 1970s-style inflation shock, high tech valuations in general, and a potential freeze in private credit markets don’t bode well for stocks.
“The entire market has become one giant AI structure,” he says. “The danger is a repeat of the dotcom bubble—a massive crash and years of lost returns. By some measures, valuations aren’t as stretched as they were back then, but this looks like an incredibly risky market.”
2. Spending is growing at an astonishing rate
Spending on AI—from data centers to chips—is racing ahead, from $765 billion this year to $1.6 trillion by 2031, according to Goldman Sachs. The investment bank acknowledges there could be problems with such a huge commitment. What if data centers are delayed?
“With the scale of capital being committed, even small delays in execution invite serious questions about the demand assumptions behind these investments,” say Goldman analysts. However, they add that if the spending plans go smoothly, it could spark a new wave of AI demand. Still, the spending shows just how much global financial resources—and expectations for returns—are being poured into AI.
3. Companies and consumers are adopting AI quickly
Despite mixed reports on its benefits, the vast majority of companies are starting to use AI—up from 33% in 2023 to nearly 80% now, according to consulting firm McKinsey. Public usage is also high, with OpenAI’s ChatGPT now reaching 1 billion monthly active users, according to data from Sensor Tower—a record for any app.
The challenge for AI developers now is how to make money from this huge base of public and private customers. Companies need to show that AI improves outcomes and cuts costs enough to justify the expense. That means using it to build entire workflows—business jargon for completing a task from start to finish. There’s still a long way to go on that front.
4. Claude is catching up to ChatGPT
Anthropic started gaining ground on OpenAI late last year, when its Claude Code tool went viral among software developers, mostly in the San Francisco area, before spreading more widely. Claude Code marked a shift in how large language models—the core technology behind chatbots—are used, moving toward autonomous AI agents that perform tasks without human help, allowing even non-tech-savvy people to create software and handle a wide range of tasks.
OpenAI still has a much larger overall user base, but data from internet analysis company Kentik—which tracks usage across several US internet service providers—shows that Anthropic is quickly closing the gap. Claude’s user traffic grew significantly faster than ChatGPT and Google’s Gemini between January and April, spiking after that period.The Pentagon labeled it a supply chain risk in March. At this rate of growth, Kentik predicts it could surpass ChatGPT by summer—another reason why Anthropic might find an easier path to an IPO than its competitor.
5. AI is getting more expensive to use
Every time an AI chatbot or agent gives a response, it’s measured in “tokens”—basic units of language that can be words, punctuation marks, or syllables. (For example, OpenAI says the phrase “You miss 100% of the shots you don’t take” is worth 11 tokens.) Tokens also measure inputs, like the prompt you type into ChatGPT.
The costs vary by model; OpenAI charges $5 per million input tokens for GPT-5.5, and $30 per million output tokens (the response to your prompt).
The problem for users is that token costs are rising sharply, even as companies everywhere push employees to “tokenmaxx”—meaning, really go all in on using AI. The problem for AI companies is that they still aren’t charging enough.
The basic promise of AI is that the money a company spends on these tools is more than made up for by gains in productivity—a measure of economic efficiency where higher productivity means more output per worker. If this trade-off isn’t working, then the assumptions behind AI valuations—and policies—are weakened.
“The costs are getting completely out of control,” says Liam Betsworth, founder of the British AI startup Pendra. He says software developers in his network are using agents to code, starting with the cheapest subscription and quickly moving to the most expensive one. They’re not alone—news site Axios recently reported on an unnamed company that spent $500 million in a month on licenses for Claude Code.
6. Data center construction might not keep up with demand
Building data centers is like the central nervous system of AI products, so growing development and use of AI tools must be matched by more capacity—otherwise there will be a compute crunch, meaning higher costs for AI companies and users.
The sector’s ambitions for data centers are huge and seem almost unrealistic. Bloomberg estimates that 23 gigawatts of capacity were under construction globally in 2025 (capacity is measured in electrical power, because that limits how much computing a site can do).
The US property company JLL predicts that 100 gigawatts will be added between 2026 and 2030—doubling what they estimate as current capacity, equivalent to 1,200 data centers. JLL says its estimate includes speculative projects that may never break ground.
Where the money—and energy supply—will come from to meet this forecast is an open question. Cecilia Rikap, an associate professor at University College London, says many projects around the world depend on political promises to expand the grid and deliver power; but governments might not have the resources to follow through.
She asks: “Has the government calculated whether such an expansion is possible? Do they have the money to do it? Have they considered the environmental damage it would cause?”
7. What AI models can do is expanding rapidly
The abilities of AI models have improved by leaps and bounds since 2023, according to METR, a research organization that measures AI capabilities.
METR’s measurements are based on whether AI models can complete a coding task, measured by how long it would take a human to do it. By this measure, AI models are doubling in capability every four months. For example, Anthropic’s Claude Mythos model is estimated to achieve a 50% success rate on tasks that would take a human expert between eight hours and two days.
However, there hasn’t been a matching impact on jobs—so far. A March report from Anthropic included research showing that, in theory, AI could perform many jobs, from computing to legal work, but it hasn’t done so on a large scale yet.
Bouke Klein Teeselink, an academic at King’s College London and an expert on the impact of AI, notes that this gap remains significant.The impact of AI on work shows there are obstacles to adopting it in the workforce. For example, how much of a CEO or senior manager’s job can safely be handed over to a bot? Can legally sensitive tasks be done by anything other than a human? Still, he says, change is coming.
“We’re still in the early stages of the AI revolution. Many people are doing tasks that could be handled by AI. The scale of change we’re about to see will be enormous.”
8. Data centers are propping up US GDP
Even though the US government has cut jobs under Donald Trump’s administration and many industries have seen mass layoffs, US GDP has kept growing—2.1% in 2025 and 1.6% in the first quarter of 2026, according to the US Bureau of Economic Analysis. However, a Harvard economist calculates that without the data center boom, these numbers could be much smaller. In fact, “investment in information processing equipment and software” made up 92% of US GDP growth in the first half of 2025.
This means that data centers—and the AI boom—are driving a huge share of US growth. They’re a big reason why the world’s largest economy still looks healthy, despite major challenges. Any slowdown in this spending could have economic, and therefore political, consequences.
Frequently Asked Questions
Here is a list of FAQs based on the article Billions have been spent and the potential returns are still uncertain Heres the AI boom explained with six charts
BeginnerLevel Questions
1 Why are companies spending so much money on AI if the returns arent guaranteed
They are betting that AI will eventually revolutionize industries and create massive profits similar to the early days of the internet They dont want to be left behind if it succeeds
2 What do you mean by AI boom
Its the current period of huge investment excitement and rapid development in artificial intelligence especially in tools like ChatGPT and image generators
3 Is AI actually making money for anyone right now
For most companies not yet The cost of building and running advanced AI is huge and its still unclear when those costs will be paid back by widespread paying customers
4 What are the six charts in the article about
They show key trends massive spending by tech giants surging demand for AI chips the huge energy cost of AI the small number of actual paying users the performance plateau of some models and the uncertain stock market reaction
AdvancedLevel Questions
5 Why is the cost of inference such a big problem
Inference is when you actually use an AI model Its far more expensive than training the model once If millions of people use it daily the electricity and computing costs skyrocket making profits very hard to achieve
6 How do the charts show that the easy gains in AI might be over
One chart shows that the performance improvements of the biggest AI models are starting to slow down even as the cost to train them keeps rising This suggests we are hitting a point of diminishing returns
7 What does the article say about Nvidias role in this boom
Nvidia makes the essential chips for AI The charts show that Nvidias revenue has exploded but this is the picks and shovels of the gold rush The question is whether the companies buying those chips will ever make a return on their investment
8 How does the uncertain return affect smaller AI startups
Its
