Baird Investment Strategist Ross Mayfield recently appeared on CNBC to push back against the prevailing narrative that the memory cycle has been permanently rewired by AI demand, framing the bull case for Micron Technology (NASDAQ:MU | MU Price Prediction) as a position-management problem rather than a definite outcome. His warning comes on a charged day, with Micron reporting fiscal Q3 2026 earnings tonight, June 24, after the market closes. The stock fell roughly 13% on Wednesday as traders reset expectations ahead of the report.
Mayfield’s core argument is that memory remains structurally cyclical, and price matters. It is a seller’s market today, but if elevated DRAM and HBM prices are expected to persist deep into 2027 and 2028, the largest buyers have both the capital and the motivation to engineer their way around Micron’s pricing power. He pointed to companies like Google and Broadcom pursuing compression software and custom-built silicon, and Amazon exploring in-house designs, as evidence that hyperscaler capex can be redirected when memory becomes a constraint rather than a commodity.
The Setup Mayfield Is Worried About
Mayfield characterized memory names as up roughly “1,000%” in a year on air. Micron closed at $1,051.77 on June 23, 2026, against $121.78 a year earlier, a 763.64% one-year move. Year-to-date, Micron is up 268.68%, and over five years, the stock has returned 1,238.55%. The market cap now sits near $1.37 trillion, with a trailing P/E around 57 and a forward multiple near 11.
In fiscal Q2 2026, Micron reported $23.86 billion in revenue, non-GAAP EPS of $12.20, and a GAAP gross margin of 74.4%, up from 36.8% a year earlier. Management guided Q3 to revenue of $33.5 billion plus or minus $750 million and a gross margin of approximately 81%. CEO Sanjay Mehrotra told investors that “In the AI era, memory has become a strategic asset for our customers” as the board approved a 30% dividend increase.
Why Margins Are Important for the Industry
Mayfield’s core point: an 81% gross margin guide is the kind of number that invites competition. Micron’s Cloud Memory segment posted a 74% gross margin and a 66% operating margin in Q2, with $7.75 billion in revenue.
Hyperscalers paying these prices have a strong incentive to fund alternatives. Google’s TPU roadmap, Amazon’s Trainium silicon, and Meta’s MTIA program already lean on architectural tricks that reduce HBM dependency per training run. Broadcom continues to ship custom ASIC programs for the same buyers. Memory demand remains intact, but this dynamic caps how long suppliers can price as if memory were uniquely scarce.
The historical pattern reinforces the caution. Just two fiscal years ago, Micron was reporting negative EPS through the 2023 trough. The recovery has been steep: $1.56 in Q3 FY2025, $4.78 in Q2 FY2026, and $12.20 in Q3 FY2026. Order books are reportedly extending into 2027, but Mayfield’s question is what 2027-2028 capacity and pricing look like once new fabs ramp and customer workarounds mature.
Key Takeaways for Micron
Mayfield’s warning is that Micron’s extraordinary profitability today may encourage the world’s largest technology companies to build alternatives. Micron’s earnings report tonight will help determine whether AI-driven demand remains powerful enough to outweigh that risk.