AI Bubble and Market Risks: Energy Supply Becomes the Key Constraint

Source: Blockworks Original Title: Friday charts: Elements of a bubble Original Link: https://blockworks.co/news/elements-of-ai-bubble

“There has been a lot of talk about an AI bubble. From our vantage point, we see something very different.”

— Jensen Huang

Alphabet CEO Sundar Pichai said there are “elements of irrationality” in the current boom in AI infrastructure.

But the announcement of a new version of Google’s Gemini LLM gave reason to think we may not be irrational enough. Gemini 3 was received as a surprisingly massive improvement over Gemini 2.5, as measured by the metrics that language models are judged on.

This disproves the “scaling wall thesis” that LLMs had hit a plateau where simply adding more compute no longer resulted in better performance. Google added better compute—smarter algorithms, better training, newer chips—and Gemini 3 got significantly better.

This seems like a green light for everyone to keep investing aggressively in everything. Jensen Huang noted that everyone is still investing heavily in GPUs: “Blackwell sales are off the charts, and cloud GPUs are sold out.”

Nvidia’s CFO added that “the A100 GPUs we shipped six years ago are still running at full utilization today.” GPUs appear to be benefiting from a “cascading use model”: The newest chips are used for training for a year or so and then run inference tasks for a while longer before eventually serving older applications.

The Energy Constraint

If the models are still getting better, demand for new chips is increasing and old chips are still useful, we should wonder whether AI companies are understating their earnings. Yet stocks were lower this week.

This might be a sign that the market has stopped worrying about the demand for chips and started worrying about the supply of energy.

Demand seems to be close to insatiable: A Google Cloud executive estimated it would have to double its compute capacity every six months for the next four or five years to meet demand.

But where the power will come from is a mystery. It takes five to seven years to build the gas turbines that power most data centers, and the companies that make them are fully booked until at least 2030.

If additional power isn’t available, there’s no point in buying a new-generation GPU that draws more of it. The AI bubble could pop even if the demand for AI is effectively unlimited.

Pickhai warned that if the AI bubble does pop, “no company is going to be immune.” Without the boom in data centers, the US economy would likely be in recession: Data centers, which account for 4% of GDP, accounted for 93% of GDP growth in the first half of the year.

Key Market Indicators

Capex as Percentage of GDP: As measured by capex as a percentage of GDP, the AI boom is already similar in scale to the investing booms that preceded the dotcom, housing and shale bubbles.

Older GPU Demand: Demand for older, less-powerful A100 GPUs has held up surprisingly well, contradicting some bearish predictions.

Corporate Capital Allocation: At Microsoft, capex has surged to nearly 50% of sales, showing a shift “from competing on network effects to competing on access to capital.” The latter is much more bubble-prone.

Startup Focus: Nearly all Y Combinator startups are now AI-related, suggesting concentrated risk in the ecosystem.

Employment and Economic Effects

From the starting point of ChatGPT, the S&P 500 has shot higher and the number of job openings in the US has shot lower. However, history suggests that technological displacement often leads to job reallocation rather than permanent loss.

The delayed jobs data showed the US with a surprise gain of 119,000 jobs in September, suggesting the labor market remains resilient despite AI disruption concerns.

Broader Economic Context

Tariffs have made imported goods approximately 5.44% more expensive than they otherwise would have been. The average price of a car in the US is now above $50,000, while in some places, gasoline is cheaper than water.

The median age of US house buyers has risen to 59, reflecting challenging homeownership economics for younger generations.