Micron Technology Inc. (MU) has become one of the cleanest ways to bet on the AI infrastructure boom. Surging memory demand, constrained supply, and heavy data-center utilization are all pushing pricing and earnings expectations higher.
The company's latest results reinforced the case for a longer AI-driven memory cycle. But analysts are still arguing about how long elevated pricing can hold up once new capacity starts coming online.
The stock has gained almost 300% in 2026, powered by an AI-led memory "supercycle" that has helped create a global supply shortage.
Reitzes Sees Memory Cycle Stretching Into 2028
Melius Research's Ben Reitzes told CNBC that Micron's outlook pushes back on fears that extra memory capacity in 2028 will automatically kill the current upcycle.
He pointed to management's view that supply-demand conditions could actually stay tighter in 2028 than in 2026, even after planned capacity additions come online.
Reitzes also expects margins to improve after the first quarter rather than fall apart as new supply hits the market.
His bigger argument: investors should start treating memory as a critical AI component instead of a traditional commodity. More memory capacity can improve AI performance and power efficiency, which changes the math.
Reitzes also sees share repurchases becoming a meaningful support for Micron, estimating that annual buybacks could eventually exceed 10% of its market capitalization.
Feeney Sees AI Capacity Opportunity but Flags Pricing Risk
Advisors Capital Management partner and portfolio manager JoAnne Feeney took a more cautious view of Micron itself, while still treating its results as another signal that the AI infrastructure buildout has further to run.
Feeney told CNBC that Micron's strong results and comments that memory demand remains above supply point to a need for additional industry capacity.
She said that dynamic could benefit semiconductor-equipment companies such as Lam Research Corp. (LRCX), which Advisors Capital owns in its growth strategy.
But Feeney cautioned that Micron has benefited significantly from sharply higher memory prices. She sees risk when those prices eventually decline and investors begin anticipating a turn in the cycle.
Feeney prefers broader AI exposure through more vertically integrated companies such as Alphabet Inc. (GOOGL) and Microsoft Corp. (MSFT), rather than relying directly on memory pricing.
Arya Says AI Demand Continues to Outpace Supply
Bank of America Securities senior semiconductor analyst Vivek Arya remains highly bullish on the broader semiconductor sector.
Arya told CNBC that AI has moved from consumer chatbots to enterprise applications and agentic workloads, with physical AI and robotics potentially creating another demand wave.
He estimates AI-related demand is growing by more than 100% annually, while semiconductor capacity can expand only about 40%-50% per year. That imbalance, he said, supports strong pricing power.
Arya considers concerns about incremental memory supply overstated because he expects underlying demand to remain substantially stronger.
He also highlighted Micron's revenue and gross-margin outlook and said memory now represents roughly half of the bill of materials for data-center systems.
High Utilization Separates This Cycle
Arya also sees a key difference between today's AI boom and previous infrastructure cycles: high utilization.
Older and newer generations of NVIDIA Corp. (NVDA) chips can serve different workloads, while newer Blackwell and Vera Rubin systems address more demanding applications.
That leaves little unused computing capacity and supports continued demand for memory and other semiconductor components.
Together, Reitzes, Feeney and Arya see strong underlying AI infrastructure demand, but their views highlight the central investor debate around Micron: whether persistent demand can continue absorbing new supply before memory pricing eventually turns.
MU Price Action: Micron Technology shares were up 0.87% at $1106.94 during premarket trading on Friday, according to market data.
