AI chip boom: Jensen Huang says this cycle differs

Key takeaways

• Nvidia chief Jensen Huang says today’s AI chip boom rests on real use, not just hope.
• Companies are spending on data centres that run AI tools for millions of people.
• Chip demand can still cool fast if spending outruns useful AI work.
• Nvidia’s past sales show why investors watch data-centre budgets closely.

The AI chip boom is a rush to buy powerful chips that train and run artificial intelligence. Nvidia CEO Jensen Huang says this surge differs from older chip manias. He argues that businesses now need more computing power for real work. That could make demand last longer, even if the ride stays bumpy.

Why does Jensen Huang think the AI chip boom is different?

Huang’s main point is simple. The chips are not sitting idle while people wait for a future idea. They power chatbots, code helpers, search tools, factory software, and research projects today.

Fortune reported that Huang expects the usual boom-and-bust cycle to look different this time. His case rests on a shift in how firms use computers. Older data centres mainly stored files and served web pages, while AI data centres create answers, images, software, and predictions.

That change matters because AI tasks need far more math than normal office apps. A data centre is a building packed with computers. Its chips, power gear, cooling systems, and network links can cost billions of dollars.

The AI chip boom can last if companies earn real value from AI tools, but it can fade quickly if costly data centres produce little useful work.

Huang has strong reasons to make this argument. Nvidia sells the graphics processing units, or GPUs, that many AI systems use. A GPU is a chip built to handle many maths jobs at once.

Still, his view is not a guarantee. Chip makers have seen sharp ups and downs before. Customers can cut orders when they own enough machines, or when money gets tight.

What do the numbers say about the AI chip boom?

Nvidia’s own results show how large the shift has become. In fiscal 2025, Nvidia reported revenue of $130.5 billion. That was up 114% from the year before, according to its annual report filed with the US Securities and Exchange Commission.

Its data-centre unit brought in $115.2 billion that year. That equals about 88% of total revenue. Five years earlier, the company’s data-centre sales were $2.98 billion.

Nvidia data-centre revenueUS dollars, fiscal yearsFY2020: $2.98bnFY2025: $115.2bn

Those figures explain why investors link Nvidia to the wider AI chip boom. Yet one company’s fast growth does not prove every chip stock will win. Nvidia has a leading software system called CUDA, which helps programmers use its chips, but rivals are working hard to catch up.

Measure	Figure	Why it matters
Nvidia FY2025 revenue	$130.5bn	Shows the scale of AI spending
Data-centre revenue	$115.2bn	Made up about 88% of sales
Year-on-year revenue growth	114%	Growth was unusually fast

What could slow the AI chip boom?

The biggest risk is overbuilding. Big cloud firms may buy more chips than their customers need at first. Cloud firms rent computing power over the internet, much like renting a bike instead of buying one.

Power is another limit. AI data centres use huge amounts of electricity, so new sites need grid links and cooling. A delayed power line can leave expensive chips waiting in a warehouse.

Prices can also fall. If more firms make capable AI chips, buyers gain choices. That could squeeze profit even if total chip use keeps rising.

Investors should also separate chip sales from stock prices. A share price can jump because people expect future growth. If those hopes are too high, the price can drop even when a company still grows.

Why should India watch the AI chip boom?

India does not yet make the most advanced AI chips at scale. But Indian firms can gain through cloud services, software, data centres, and chip design. More AI use could also raise demand for reliable power and fast networks.

For local investors, the story reaches beyond one US company. Memory makers are important because AI servers need lots of fast memory. That link was clear in the Nvidia and SK Hynix memory deal.

India’s policy choices matter too. Rules for data, electricity, and skills will shape where new AI capacity gets built. Readers can also see why clear rules are being sought in our report on the push for an India crypto policy.

How should readers judge AI chip claims?

Watch whether AI tools save time or bring in money. A company that uses AI to answer customer questions faster has a clear reason to keep paying. A company buying chips without a plan may stop later.

Also watch capital spending. Capital spending means money used to buy long-lasting things, such as servers and buildings. Nvidia posts detailed quarterly results and forecasts on its investor relations page.

The AI chip boom has real fuel because AI needs physical machines. But Huang’s argument will face a hard test: can customers turn all that computing power into steady value?

FAQs

What is the AI chip boom?

The AI chip boom is rising demand for chips that run artificial intelligence. These chips sit in large data centres and process huge amounts of data.

Why are Nvidia chips so popular for AI?

Nvidia chips are fast at the repeated maths used by AI. Its CUDA software also helps developers run AI programs on those chips.

How could the AI chip boom end?

The AI chip boom could cool if firms spend too much before AI earns enough money. More competition, power limits, or weaker demand could also reduce orders.