MRAM · Everspin Technologies, Inc. · Stock research

Current Thesis

The post-parabolic markdown stalled rather than reversed. After running roughly +300% into a mid-$40s all-time high in early May and then marking down ~50%, Everspin found a floor in the low $20s on the 50-day moving average through June. Two exogenous inputs built that shelf: Micron's blowout fiscal Q3 print (2026-06-24) reignited the memory-supercycle narrative, and Russell 2000 inclusion (effective the 2026-06-26 close) added a slug of passive demand. Neither is company-specific. Everspin is the niche, most-shorted, weakest-exposed name in the memory basket — industrial and embedded persistence memory, not the DRAM/HBM that is actually short — and at the last confirmed print of $21.75 (2026-06-26) it traded ~17% above the $18 average sell-side target on a P/E near 1,800. The Q2 report, expected ~2026-07-28, converts a sympathy-bid tape into a binary: the company either shows a non-casino datacenter or AI design win that justifies the multiple, or the markdown toward the $14–18 target cluster resumes. Entering a 1,800x-earnings microcap into a print it has never de-risked, while an open short campaign targets $14, is a stand-aside.

Bullish and bearish views on Everspin Technologies, Inc.

The model's bull view on Everspin Technologies, Inc. (MRAM), in brief: Memory supercycle intact: Micron's fiscal Q3 2026 print (2026-06-24) was strong enough that shortage is bleeding into consumer hardware pricing (Apple laptop and Microsoft Xbox hikes cited 2026-06-24/25); Benzinga tagged MRAM into the "memory rally restarts" cohort on 2026-06-25. The bear view: At the last confirmed $21.75 (2026-06-26) the stock sat ~17% above the $18 average analyst target (2 analysts) and far above Kerrisdale Capital's $14 short objective (disclosed 2026-05-19), which remains open. Both cases follow in full.

Bull Case

• Memory supercycle intact: Micron's fiscal Q3 2026 print (2026-06-24) was strong enough that shortage is bleeding into consumer hardware pricing (Apple laptop and Microsoft Xbox hikes cited 2026-06-24/25); Benzinga tagged MRAM into the "memory rally restarts" cohort on 2026-06-25.
• Russell 2000 inclusion effective at the 2026-06-26 close (announced 2026-06-09; market cap ~$510M) added passive index demand and broader institutional visibility.
• Q1 2026 (reported 2026-04-29): revenue $14.9M, +14% YoY, beat the $14.63M estimate; MRAM product sales $14.1M, +28% YoY; gross margin 52.7%; adjusted net income $2.6M versus a year-ago loss.
• Contracted revenue base: a 10-year supply agreement with Microchip plus a $40M, 2.5-year Toggle-MRAM defense subcontract into the US Defense Industrial Base, part of which booked in Q1.
• Price defended the 50-day MA in the low $20s for roughly three weeks (early-to-late June); Benzinga (2026-06-25) flagged a MACD reversal off that support as the technical trigger.
• Thin float plus persistent retail engagement keeps squeeze optionality live on any positive memory or design-win headline into the ~2026-07-28 print.

Bear Case

• At the last confirmed $21.75 (2026-06-26) the stock sat ~17% above the $18 average analyst target (2 analysts) and far above Kerrisdale Capital's $14 short objective (disclosed 2026-05-19), which remains open.
• Valuation is extreme: P/E ~1,795 (2026-06-26) against revenue anchored near a $60M annual run-rate for years; +14% YoY does not underwrite a triple-digit-percent re-rate.
• The supercycle being borrowed is a DRAM/HBM story. Everspin sells industrial and embedded persistence memory with no hyperscaler demand exposure; Kerrisdale's 2026-05-19 report identifies casino gaming and slot machines as the largest end market.
• Relative weakness is the tell: memory peers broke out on the 2026-06-24 Micron print while MRAM held ~50% below its early-May high — the basket ran without it.
• Insider distribution is one-directional: ~$3.9M sold in the prior quarter with zero buys, then Director Lawrence Finch's 310,091 shares / $8.52M (~$27.47) on 2026-06-02, straight into the squeeze.
• Headline Q1 EPS of $0.40 was flattered by ~$2.1M of one-time defense other-income; the underlying result was roughly a $0.3M net loss — the earnings power the multiple prices does not exist yet.
• Russell-add demand is exhausted at the effective date; index-inclusion pops frequently fade once passive buyers are filled.

Setup & Price Structure

• 52-week range $5.76–$51.50. The Q1-driven parabola ran ~+300% to a mid-$40s all-time high in early May, then marked down ~50% into the low $20s.
• Last confirmed tape: ~$29.25 (2026-06-04) → ~$23.05 (2026-06-06) → $22.55 (2026-06-23) → $21.75 (2026-06-26). This dossier's price reference is 21 days stale as of 2026-07-18 and must be re-validated against a live quote before acting.
• Structure is a post-parabola base sitting on the 50-day MA in the low $20s. The shelf that matters is ~$20; losing it on a weekly basis forfeits the base and reopens the analyst-target cluster ($14–18) as air beneath.
• The name carries a high short interest and a thin float, so tape is bimodal — it either squeezes on a memory headline or bleeds on distribution days. That is a trader's vehicle, not an investor's, and it argues for small size and a hard level rather than conviction.
• Into the ~2026-07-28 print, an accelerating-momentum entry does not exist here: the stock is digesting, not breaking out, and the binary event caps upside asymmetry for a fresh long.

Catalyst Calendar (next 30 days)

No upcoming dated catalysts on file — the dated entries below have passed.

Elapsed catalysts

• ~2026-07-28 (est.): Q2 2026 earnings print. The dominant binary. Watch for (a) any datacenter/AI design win outside casino gaming, (b) guidance above the ~$60M run-rate, (c) gross-margin trajectory versus the 52.7% Q1 mark. Q1 was reported 2026-04-29, supporting the late-July cadence. (passed 12d ago)
• ~2026-07-23 (est.): earnings blackout begins ~3 trading days ahead of the print — fresh entries into the event are a pass. (passed 17d ago)
• Ongoing: Kerrisdale Capital's $14 short thesis (2026-05-19) is unresolved; a public cover or an escalation is a live tape driver either direction. (passed 82d ago)
• Ongoing: any Form 4 activity following the 2026-06-02 Finch sale — a first insider buy would be a genuine signal change; continued selling confirms distribution. (passed 68d ago)

What Would Change Our Mind

• Bull flip: a Q2 print (~2026-07-28) that discloses a named non-casino datacenter or AI-adjacent design win, plus guidance breaking the $60M run-rate, would convert the sympathy bid into a fundamental re-rate and justify the multiple the market is already paying. A weekly close reclaiming the mid-$20s on expanding volume, alongside a Kerrisdale cover, would confirm.
• Bear confirmation: a weekly close below $20 forfeits the low-$20s base and the 50-day-MA shelf, reopening the $14–18 analyst cluster; a Q2 report that reaffirms the ~$60M run-rate with no design win removes the only thing the bulls can point to beyond memory sympathy.
• The swing factor is exposure, not price: MRAM re-rates only if it stops being a casino-memory supplier borrowing a DRAM narrative and becomes an actual participant in the shortage. Absent that, every memory headline is a rent, not a durable bid.

Correlation Notes

• Trades as high-beta sympathy to the memory complex — Micron (MU), the NAND names, and small-cap peers such as GSI Technology (GSIT), which appeared alongside MRAM in the same 2026-05-11 gainer lists. The correlation is directional, not fundamental: MRAM lacks the DRAM/HBM demand that drives the leaders.
• Tied to Russell 2000 / IWM passive flows post the 2026-06-26 inclusion; small-cap risk-appetite swings move it more than company news between prints.
• Sits in the retail-squeeze microcap cohort (thin float, high short interest, active short-seller battle), so it co-moves with meme-flow risk sentiment on high-volume days independent of the memory tape.
• Decouples from the basket to the downside when memory risk-off hits, because it is the marginal, most-shorted holding investors sell first — the mirror image of why it lagged the 2026-06-24 breakout.