One of many first market truths I discovered over my 30 years navigating Wall Road was that semiconductor demand is cyclical. When occasions are good, orders and costs surge, driving suppliers to ramp up capability too rapidly, which contributes to oversupply when demand softens. Rinse. Repeat.
Get the timing of these cycles right, and traders could make a major amount of cash, significantly in reminiscence makers like Micron, which is traditionally liable to the growth bust cycle.
Whereas no one rings a bell signaling the official begin and finish of demand supercycles, those that take note of issues like spot market pricing are more and more beginning to pound the desk that we could also be on the cusp of a crucial inflection level for Micron (MU), together with Wall Road analysis agency Stifel.
This week, Stifel analyst Brian Chin put the scenario at Micron bluntly, writing in a analysis report shared with TheStreet:
Chin notes that surging spending on knowledge facilities to outfit them with {hardware} essential for dealing with synthetic intelligence workloads has confirmed a “tipping level” for the reminiscence market, inflicting costs to swell, and in flip, main him to ratchet his Micron inventory worth goal up by an eye-popping 54% forward of its earnings subsequent week.
His larger goal is the newest amongst a slate of latest Wall Road friends, suggesting many are behind the eight ball and speeding to catch up.
Reminiscence costs are surging in 2025 on account of rising demand for synthetic intelligence servers.Shutterstock&interval;
Tons of of billions have been spent over the previous three years upgrading knowledge middle servers to raised deal with the heavy workloads related to coaching and operating giant language fashions, similar to ChatGPT, and agentic AI apps tasked with boosting employee productiveness.
In 2025 alone, I/OFund says hyperscalers capex can be $405 billion.
The frenzy of exercise has been a boon for Nvidia, the semiconductor firm behind the chips handiest at AI, and server firms, like Dell, that are constructing extra highly effective computer systems filled with Nvidia’s GPUs to satisfy hyperscaler demand.
Extra Tech Shares:
“Coaching is considerably and more and more compute-intensive, however early LLM calls for have been manageable. Right now, compute wants are accelerating quickly, significantly as extra fashions transfer into manufacturing,” wrote JP Morgan strategist Stephanie Aliaga in October. “Nvidia estimates that reasoning fashions answering difficult queries might require over 100 occasions extra compute in comparison with single-shot inference.”
The sheer horsepower essential to meet AI’s mission is now spreading past early pick-and-shovel performs, like Nvidia and Dell, to interconnect merchandise used to hyperlink programs collectively and reminiscence used to make AI servers run extra effectively.
In its newest earnings name, Dell acknowledged the affect of rising demand on reminiscence, citing the reminiscence market as an rising ache level.
“We’re in a really distinctive time. It’s unprecedented. We have now not seen prices transfer on the fee that we’ve seen. And by the way in which, it’s not distinctive to DRAM. It’s NAND,” stated Dell Vice Chairman Jeffrey Clark.
The spot worth market has seen reminiscence costs surge larger because the summer time. As these costs are mirrored in contract renegotiations, it can provide a major bottom-line-friendly revenue for Micron.
Stifel’s Chin up to date his outlook for ahead gross margins and earnings, and the scale of the rise, which is anticipated to speed up in every of the following two quarters, was sufficient to justify a serious enhance in Micron’s inventory worth goal to $300 from $195.
“Reminiscence costs are rallying laborious into year-end. Subsequent week, we anticipate Micron to report and information for significant upside to its F1Q(Nov) outcomes and F2Q(Feb) outlook,” wrote Chin. “Within the brief run, working at larger utilization and decrease stock, output is extra fastened and the priority for Micron is just not about an excessive amount of provide however moderately too little.”
That is a superb downside to have in the event you’re Micron. Stifel believes that Micron will exceed its quarterly outcomes when it studies on December 17.
It expects income and non-GAAP earnings per share of $13.1 billion and $4.12, respectively. That is properly above Wall Road consensus for $12.8 billion and $3.91, and far larger than Micron’s personal steering for not less than $12.2 billion and $3.60.
The revenue outperformance can be pushed by gross margins increasing 7.8% quarter over quarter to 53.5%, which is above the excessive finish of its steering of fifty.5% to 52.5%. The outcomes mannequin for 15% sequential gross sales progress, with worth being the main driver, moderately than unit quantity.
Importantly, Chin believes that Micron will acknowledge that “contract pricing is broadly strengthening into year-end,” offering upside for its outlook for the present quarter.
For the upcoming fiscal second quarter, Stifel thinks the corporate can ship $14.9 billion in gross sales and $5.09 in EPS. Once more, that is higher than Wall Road consensus, which is pegged at $13.9 billion and $4.56.
Gross margin might enhance one other 4% from fiscal Q1 to 57.5% and additional out, Stifel’s analysts write, “we consider Micron Cloud Reminiscence (CMBU) GMs might increase into the mid-60s (59% in F4Q).”
The outlook suggests compelling sequential top- and bottom-line progress, an ideal recipe for investor features, provided that shares are likely to observe income and revenue over time.
Stifel is way from the one agency boosting targets and touting a bullish tone. Goldman Sachs’ analysts anticipate Micron to ship third-quarter income of $13.2 billion and earnings of $4.15 per share as “pricing energy ought to drive upside to quarter, positioning the corporate properly for 2026.”
Sustainable Progress Advisers (SGA), an funding administration firm, launched its fourth-quarter investor letter for its “U.S. Massive Cap Progress Technique.” A duplicate…
