The semiconductor industry is the beating heart of the artificial intelligence (AI) revolution, because most development happens in large data centers that are filled with thousands of graphics processing units (GPUs) from chipmakers like Nvidia (NVDA 0.87%) and Advanced Micro Devices (AMD -0.20%).
Micron Technology (MU 1.22%) is another leading chip company, but it doesn't supply GPUs, so it receives less attention than the likes of Nvidia. Instead, Micron makes memory and storage chips, which are becoming extremely important for processing AI workloads, not only in data centers, but also in computers and smartphones.
Micron is scheduled to release its financial results for its fiscal 2025 third quarter (which ended on May 30) on June 25, and the company's guidance suggests the report could reveal substantial growth at the top and bottom line, driven by AI. Should investors buy the stock ahead of the release?
Image source: Getty Images.
Memory is a key ingredient in AI workloads
GPUs are ideal for AI workloads because they are designed for parallel processing, meaning they can handle multiple tasks at once with a high degree of efficiency. But GPUs need memory to store data in a ready state, so it can be called upon at a moment's notice to speed up processing time.
Micron's HBM3E (high-bandwidth memory) for the data center is the best in the industry. In fact, Nvidia selected it to power its latest Blackwell and Blackwell Ultra GPUs, which are designed for next-generation "reasoning" AI models. These models require significantly more computing power and memory than traditional large language models (LLMs), because they spend more time thinking in the background to craft the most accurate responses for users.
Micron expects the market for data center HBM to double to $35 billion this year, and then grow to $100 billion by 2030. The company is already completely sold out of HBM3E until 2026, and it's now working on HBM4E to stay ahead of the competition. This new solution will launch next year, and it could deliver a whopping 60% more bandwidth than its predecessor.
But memory is also a critical component in computers and smartphones -- even more so now that some smaller AI workloads are designed to be processed on-device. In other words, chips are becoming powerful enough to run some AI applications (like chatbots) offline, so they don't need to lean on any computing power from external data centers.
As a result, Micron says the minimum DRAM (memory) requirement in an AI personal computer is now 16 gigabytes, up from 12 gigabytes last year for non-AI computers. AI-enabled smartphones have also experienced a jump in memory demand to 12 gigabytes, compared to 8 gigabytes for their non-AI counterparts.
Micron's revenue likely soared in the third quarter
During the fiscal 2025 second quarter (ended Feb. 27), Micron generated $8.1 billion in total revenue, which was a 38% increase from the year-ago period. That included $4.6 billion in compute and networking revenue -- where the company accounts for its data center memory sales -- which was up by a whopping 109%.
Micron likely had another very strong three-month stretch in the third quarter of fiscal 2025, the results of which will be announced on June 25. The company forecast around $8.8 billion in revenue, which would be a 29% increase from the year-ago quarter. Management didn't offer any specific guidance for the compute and networking segment, but it has been the main driver of Micron's consistent strength over the past year, and that was likely the case again in Q3.
Micron's surging revenue is also driving strong growth at its bottom line. The company delivered $1.56 in non-GAAP (generally accepted accounting principles) earnings per share (EPS) during the second quarter, a 271% increase from the year-ago period. Management's guidance suggests non-GAAP EPS came in at around $1.37 in Q3, which would be up 121% year over year.
Should you buy Micron stock before June 25?
Based on Micron's $5.15 in trailing-12-month EPS, its stock is trading at a price-to-earnings (P/E) ratio of just 23.3. That makes it much cheaper than Nvidia stock, which trades at a P/E ratio of 45.7.
Since Micron's HBM3E is embedded into Nvidia's most powerful GPUs, investors who believe Nvidia's AI chip sales will continue to soar should also hold a positive view of Micron's business from here. Back in March, Nvidia CEO Jensen Huang said his company already had orders for 3.6 million Blackwell GPUs from just four customers alone, which suggests Micron's data center business could boom for the foreseeable future.
Therefore, Micron's success won't hinge on any single quarter. I think the stock could be a great buy right now based on its valuation and the demand pipeline for its memory chips, irrespective of the upcoming June 25 report. As long as investors are willing to hold the stock for the long term, they give themselves a good chance to earn a positive return.
