Nvidia and Advanced Micro Devices supply some of the world's most sought-after graphics processing units (GPUs) for data centers, which are the primary chips used in artificial intelligence (AI) development. As a result, when Wall Street talks about semiconductor companies, those two names get the most attention.
But without high-bandwidth memory (HBM), GPUs wouldn't be nearly as powerful, and Micron Technology (MU 4.83%) is one of the world's top suppliers of this hardware. In fact, both Nvidia and AMD are now using Micron's HBM in their latest AI GPUs, which is driving a surge in the company's revenue.
On Wednesday, Micron released its financial results for the fiscal 2025 fourth quarter (ended Aug. 28), and they obliterated expectations. Its stock looks like a bargain relative to Nvidia and AMD, so here's why it could be the ultimate buy heading into 2026.

Image source: Getty Images.
Memory is critical in AI workloads
GPUs are designed for parallel processing, meaning they can perform several computations at once. Memory stores data in a ready state for when the GPU needs it, which accelerates processing speeds. A low memory capacity can cause bottlenecks, forcing the GPU to constantly pause each workload, whereas a high memory capacity keeps data flowing, unleashing the GPU's maximum performance.
Micron's HBM3E solution offers 50% more capacity than the competition, and it also consumes 30% less power, which can significantly reduce costs for data center operators. That's why it was selected for Nvidia's latest Blackwell Ultra GPUs, and AMD's new MI350 Series GPUs.
Micron's entire HMB3E supply for calendar year 2026 is almost completely sold out already, and the company is now sending out samples of its newer HBM4 solution, which reportedly offers up to 60% more capacity while consuming 20% less power.
But some AI workloads are gradually migrating to personal computers (PCs) and smartphones, so these devices also require more memory capacity. Micron is a major player in both of these segments, too. The average smartphone has a memory capacity of 8 gigabytes today, but Micron says many manufacturers are now requiring 12 gigabytes or more, primarily so their devices can run AI software more efficiently. There is a similar trend in the PC market.
Higher memory capacities in these devices translates to more revenue for Micron, so this will be a massive tailwind for the company over the next few years.
Micron obliterated expectations in the recent quarter
Micron generated a record $11.3 billion in total revenue during its fiscal 2025 fourth quarter, crushing management's forecast of $10.7 billion. It represented a year-over-year increase of 45%, marking an acceleration from the 36% growth the company delivered in the third quarter just three months earlier.
But the real story is beneath the surface of the headline number. Micron's Cloud Memory Business Unit -- which is where it accounts for sales of its data center HBM -- contributed $4.5 billion in revenue during the quarter, which was a staggering 214% increase compared to the year-ago period. This highlights the significant demand for memory from AI customers.
Micron's surging revenue growth also led to a spectacular result at the bottom line during Q4. The company's earnings per share (EPS) came in at $2.83 on a generally accepted accounting principles (GAAP) basis, which was far above management's forecast of $2.29. It represented a whopping 258% jump from the year-ago result.
The good news didn't stop there, because Micron issued a very bullish forecast for the upcoming first quarter of fiscal 2026 (which concludes at the end of November). The company expects to generate $12.5 billion in revenue and $3.56 in EPS, representing year-over-year growth rates of 43% and 113%, respectively.
Micron stock trades at an attractive valuation
Micron generated earnings of $7.59 per share during the whole of fiscal 2025, placing its stock at a price-to-earnings (P/E) ratio of just 22. That makes it substantially cheaper than both Nvidia and AMD, which trade at P/E ratios of 50 and 92, respectively.
Since Nvidia and AMD are both using Micron's HBM solutions, investors who believe those companies will continue to sell truckloads of AI GPUs should also be very bullish on Micron's business. Nvidia CEO Jensen Huang predicts data center operators will spend up to $4 trillion upgrading their infrastructure to meet demand from AI developers over the next five years, so GPU sales probably won't be slowing down any time soon.
Given Micron's attractive valuation, it could be a great AI semiconductor stock to buy now and hold into 2026.