BofA Says Micron Trades at Less Than 10x Earnings and Wall Street Is Missing the Story Completely

Quick Learning

• Micron is trading at 9x forward earnings after 7 straight EPS beats, as Arya argues that memory has changed forever from a cyclical chip to an important AI.

• NVIDIA trades at 23x and Broadcom at 34x forward earnings, making Micron the cheapest entry into the same trade for the world’s AI growth.

• Arya’s bull case is set on HBM requiring 3-4x more wafers and multi-year 16-year customer agreements with low prices that stabilize the old cycle.

• Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks – and Micron Technology didn’t make it. Pick up FREE words today.

Memory stocks have recently moved out of cyclical chip games into AI-era fundamentals, and Bank of America’s Vivek Arya thinks the market is still pricing the old version. Speaking on CNBC the morning after Micron technology (NASDAQ:MU) reported, BofA’s top semiconductor analyst said “what we’re seeing is what I would call a structural rather than cyclical shift in the memory industry. It’s very important for AI. There’s no AI without memory.”

The setup behind that call is the print itself. Micron posted Q3 revenue of $41.46 billion, beating consensus by 17.60%, and non-GAAP EPS of $25.11 compared to the average of $20.28, its seventh consecutive EPS beat. GAAP gross margin fell to 84.6%, from 37.7% a year ago, a more common change for a software company than a DRAM maker. Q4 guidance calls for revenue of $50 billion and non-GAAP EPS of $31, both sequentially higher, as detailed in the company’s 8-K filing.

Take action now: the analyst who called NVIDIA in 2010 recently named his top 10 AI stocks — and Micron Technology didn’t make the cut. Pick up FREE words today.

What Arya thinks Wall Street doesn’t exist

The case of Arya’s building rests on three pillars. First, the supply is really hard to add. “You need 3 to 4 times the number of wafers to create the same amount of capacity with high-bandwidth memory as you would with a conventional product. So it’s very difficult to create.” Second, memory is becoming a reasonable chunk of a hyperscaler’s budget. Memory is about 5 to 40% of cloud capital spending, and customers put up with the price because the new computing stack pays for it. Third, agreements. CEO Sanjay Mehrotra revealed Multi-Year Strategic Customer Agreements covering 16 customers at lower prices, a structural change from the market whiplash that characterized previous cycles.

Arya is anchoring a supply-demand imbalance that lasts until the end of next year, with AI builds running at least until the end of this decade. Mehrotra’s frame tracks closely. “Micron’s record Q3 financial results and even stronger Q4 outlook demonstrate the strategic value of memory in the AI ​​era.”