Micron's long-term commitments cover only half a quarter of the revenue it now guides to

Set the commitments against the guidance and they cover about 52% of a single quarter's expected revenue ^[1]. If the guided pace simply held flat for a year, it would come to roughly $246 billion ^[2], and the commitments would be about 13% of that ^[3]. The source does not say over what period the $32 billion runs or who is committed to whom, so 13% is the generous reading. It assumes every committed dollar lands in the next twelve months. Most of Micron's high-bandwidth memory, DRAM and SSD sales ^[16] will still be priced in a market that investors expect to loosen once the HBM shortage eases, according to cryptobriefing ^[12].

Broadcom's premium rests on the opposite structure. The comparison comes from cryptobriefing's write-up of a question The Motley Fool put to investors about which name is the better AI buy over three years ^[17]. Broadcom builds custom chips for customers including Google, Meta, OpenAI and Anthropic ^[8]. Those parts are harder to swap out, and its multi-year hyperscaler partnerships give visibility that spot memory pricing cannot, the piece says ^[13]. Pair the ends of the two forward multiple ranges and Broadcom costs between 1.4 and 5 times as much per dollar of expected earnings ^[4]. At the midpoints, 9.5 against 24, it is about 2.5 times ^[5].

Micron's growth came late in the year. The fourth quarter was about 41% of fiscal 2026 revenue ^[6]. The three quarters before it averaged about $26.3 billion each ^[7], and the same quarter a year earlier was roughly $11.3 billion ^[8]. The guidance midpoint implies about 13% sequential growth, and the $1.5 billion band either side spans roughly 11% to 16% ^[9].

The bull case says memory is the bottleneck. Its evidence is a supply forecast: research findings on the company expect tight conditions through 2028 ^[6]. If that holds, earnings at this run-rate last long enough that the low multiple turns out too cautious, which is the case cryptobriefing makes ^[12]. If supply catches up during fiscal 2027, the multiple was a fair price for peak earnings. The first thing to break would be the sequential growth Micron expects through that year ^[5]. Or the gap closes from Broadcom's side. Its fiscal 2028 AI target is double its fiscal 2027 one ^[10], and the 2027 figure is itself about 1.7 times the roughly $67 billion annualised pace of its fiscal third quarter ^[11][12]. The piece notes that stocks at higher multiples tend to react sharply when growth slows ^[14].

I think the market is pricing the right variable, perhaps too harshly. Micron's annualised guidance already exceeds Broadcom's fiscal 2028 AI target ^[13], and the lower multiple attaches to the larger revenue line with less of it under commitment. Analysts rate both stocks Strong Buy ^[7][11], with about 60% price-target upside on Micron as of late September figures ^[7], so the consensus does not choose between them.

For Micron to be mispriced, its commitments have to grow faster than its sales. If the $32 billion climbs toward a full quarter of revenue while sequential growth holds, 6 to 13 times is too low. If revenue keeps rising and the commitments stall, the stock is cheap for the reason cryptobriefing gives: a shortage investors expect to end ^[12].