Everspin now looks somewhat overvalued, because its monopoly in discrete MRAM does not justify current valuations. x.com 

Credo's MicroVCSEL offers upside that does not appear priced in. Its copper and optics exposure, a SpaceX tailwind and fundamentals that look too good for its current valuation make it one of the tidiest AI infrastructure investments. x.com 

Everspin Technologies looks cheap against its sales, trading at 7x sales and 5x next year's when chipmakers usually get 6x. At 6x next year's sales, give or take 12%, fair value is $20.03-$25.49, versus $18.71 now. This is a levels-only read off reported earnings and revenue estimates, with no fundamental work. x.com 

Everspin's MRAM rally makes little sense, as the market is not that large and there is no MRAM monopoly. Absent an unknown breakthrough using MTJs for AI, there is no reason for the bid, and the space angle, raised by @ryandonovan23, fails because data centers will never be in space. x.com  x.com  x.com 

Semiconductors could capture scarce spot compute's value if its forward curve holds. Compute trades in deep contango, while chip stocks price in little of that value; near-term moves are unpredictable. That makes the sector very bullish. The earlier lab-funding risk looks less acute: Anthropic should hold nearly $200B cash after its IPO and is operating-cash-flow profitable. Even with 2029 capex cuts, chips look cheap if future buyers do not default. x.com  x.com  x.com  x.com  x.com 

Everspin Technologies is a unique memory supercycle play alongside Micron and SanDisk, as the only public standalone pure-play on MRAM, which pairs SRAM's speed with keeping data without power. TSMC and Samsung offer MRAM only as an embedded feature. Everspin has similar beta to Micron and SanDisk on a much smaller market cap, so it should swing harder, and it just had phenomenal earnings. The weekly chart shows strong support; the last buy signal brought a roughly 450% rally. x.com  x.com 

Everspin's 38.6% surge to $18.28 after its Q1 beat has likely pulled forward much of the near-term upside: at about $425M, or 7x trailing sales, it already prices in much of 2026-2027 growth. Product sales grew 28% to $14.1M and a $40M defense contract bridges about $16M a year, but reaching $25-$30 needs quarterly revenue above $20M, hinging on July 2026 STT-MRAM qualification and a Microchip ramp with recurring orders not due until 2H 2027. Litigation costs weigh on GAAP income. x.com 

Everspin's execution does not match its monopoly: revenue growth has been inconsistent and margins all over the place, so results look worse than its positioning warrants. It is the world's only at-scale maker of Toggle MRAM, radiation-immune memory the US military uses in missiles, submarines and aerospace. Its new $40M Department of War contract requires it to hand over its process technology if it ever stops making MRAM, and a Microchip partnership adds domestic production in Oregon. x.com 

Everspin is the only US-domiciled, US-manufactured MRAM maker, and its non-volatile, fast, radiation-hardened memory works where flash and DRAM cannot, making it a Pentagon supplier with no alternative. Its 10-year Microchip deal for a copy-exact line in Oregon doubles domestic capacity from H2 2027. Its UNISYST NOR flash replacement targets 5-10% of a $3 billion market, with production in 2027. It had 238 design wins in 2025, up 34%, with 52.7% gross margins and no debt. x.com  x.com  x.com  x.com  … Show more

Financing risk makes AI-linked growth stocks unattractive to add here. Much of the money reaching suppliers such as Nvidia and Micron is financed, so harder funding would limit spending on chips, compute and memory. The buildout is not over, but stock prices could remain suppressed until it slows. A significant drawdown would make adding appealing; for many high-beta names with weak fundamentals and balance sheets, the highs look to be in. x.com