TSMC and Micron earnings show AI chip demand outpacing supply

Two earnings calls, six weeks apart, describe the same shortage from opposite ends of the supply chain. TSMC sells the logic chips that train and run AI models. Micron sells the memory that feeds them. Both say demand is outrunning what they can physically produce. Both are spending heavily to close a gap they do not expect to close soon.

Start with TSMC. On Thursday, the Taiwanese foundry reported net income of NT$505.7 billion, about $16 billion, for the fourth quarter. That is up 35 percent year over year and above analyst expectations. Revenue reached $33.7 billion, a 25.5 percent increase from the same period a year earlier, according to Ars Technica. The company expects nearly 30 percent revenue growth in 2026 and plans capital expenditures of between $52 billion and $56 billion this year, up from $40.9 billion in 2025.

On the call, CEO C.C. Wei told investors he cannot predict the semiconductor industry's long-term trajectory, but he remains bullish on AI. He described demand in terms that go past the usual cautious guidance language. "All in all, I believe in my point of view, the AI is real, not only real, it's starting to grow into our daily life. And we believe that is kind of, we call it AI megatrend, we certainly would believe that," Wei said, per Ars Technica. Asked whether the semiconductor industry can stay healthy for three, four or five years in a row, he answered: "I'll tell you the truth, I don't know. But I look at the AI, it looks like it's going to be like an endless, I mean, that for many years to come."

Why the spending number matters

TSMC manufactures chips for Apple, Nvidia, AMD and Qualcomm, and produces the vast majority of the world's most advanced semiconductors. When it lifts capital spending, it is effectively converting its customers' forecasts into fab construction. Wei said he spoke directly to cloud providers to verify demand before committing. "I want to make sure that my customers' demand are real. So I talked to those cloud service providers, all of them," he said. "The answer is that I'm quite satisfied with the answer. Actually, they show me the evidence that the AI really helps their business."

That verification effort sits against months of public worry about an AI bubble. In November, Google CEO Sundar Pichai warned of "irrationality" in the AI market and said no company would be immune if a bubble bursts. OpenAI's Sam Altman acknowledged in August that investors are "overexcited" and that "someone" will lose a "phenomenal amount of money." Both remarks are cited in Ars Technica's report on the TSMC call. The earnings landed the same day the US and Taiwan finalized a trade agreement cutting tariffs on Taiwanese goods to 15 percent from 20 percent. The deal commits Taiwanese companies to $250 billion in direct US investment, and TSMC is accelerating expansion of its Arizona fabrication facilities to match. That detail matters for anyone trying to read the capex figure: part of the spending is a response to policy, not only to demand.

Micron's version of the same story

Micron's first earnings call since killing its Crucial consumer brand, held on 20 December, per Tom's Hardware, produced record revenue of $13.64 billion in its first fiscal quarter of 2026, up nearly 57 percent year over year, with what the company called significant margin expansion. Micron attributes the growth to higher pricing and increased demand in AI data centers.

Then came the less comfortable part. CEO Sanjay Mehrotra said supply constraints will persist beyond calendar 2026 and that the company is working on multi-year supply commitments. Even with capacity expansion under way, Micron expects to meet only half to two-thirds of demand from its key customers. Mehrotra said those customers are "concerned about long-term access to memory" and are lining up for multi-year contracts to secure supply.

One product explains much of the squeeze. HBM, the high-bandwidth memory used alongside AI accelerators, requires three times as much wafer space as DDR5. Micron expects strong growth in HBM revenue and projects the HBM total addressable market will reach $100 billion by 2028, which would outpace the entire DRAM market in calendar 2024. The consumer side absorbs the consequences. The DRAM shortage has pushed DDR5 prices sharply higher. Most suppliers agree the shortage will continue into at least next year and likely beyond, though they disagree on when pricing levels off. GPU vendor Sapphire says prices will stabilize in the next six to eight months, while Kingston says prices will "continue to go up," according to Tom's Hardware. Micron said it is "disappointed" it cannot meet demand across all market segments.

Capacity arrives slowly

Micron's expansion timeline shows why no quick fix exists. The company is building two fabs in Idaho, with the first expected to start producing chips in mid-2027. A New York fab is also planned: Micron expects to break ground in early 2026, with production starting around 2030. Those are multi-year horizons set against demand that is being measured in quarters.

Set the two calls side by side and the shape of the cycle becomes clearer. TSMC is raising capex from $40.9 billion to as much as $56 billion on the strength of cloud provider assurances it says it checked itself. Micron is booking record revenue while telling customers it will fall short of what they want. Neither company is describing a market that cools on its own. The open question is not whether AI demand exists today. Both companies say it does, and both are putting money behind that claim. The question is what happens to $52 billion to $56 billion of annual fab spending and a planned $100 billion HBM market if the demand curve bends before the capacity lands. Wei's own answer to the multi-year question was that he does not know. He is building as if the answer is yes.