A $40 million contract pushed MRAM to an all-time high

NEXT PICK · Market Insights

A $40 million contract pushed MRAM to an all-time high

Friday, May 8, 2026

During today's U.S. trading session, Everspin Technologies (NASDAQ: MRAM) surged from a previous close of $26.99 to an intraday high of $39.6, with a single-day gain of +46.8% and a two-day increase of 70+%, instantly expanding its market value to $927 million, pushing its price-to-earnings ratio to an almost absurd level of 4,000 times.

There is only one catalyst

On April 30, it announced a $40 million, two-and-a-half-year Toggle MRAM subcontract agreement with U.S. defense industry contractor Amentum Services.

Interestingly, on the eve of this feast, the CEO and CFO each completed a textbook share reduction. On one side is record-breaking contracts, on the other is executives selling near historic highs. Is this a fundamental turning point, or is it an emotional premium rewarded by narrative excess?

Latest price $26.99 ▲ +25.48%

Static P/E ratio 4000 times ▼ The industry median is about 25 times

Analyst mean target $18.00 ▼ Implied -33% from current price

A small factory that sells "non-volatile loss" to defense and satellite companies

Everspin is headquartered in Chandler, Arizona, and was spun off from Freescale in 2008 to develop reluctance random-access memory MRAM—a special type of memory that doesn't lose data when powered off, with read/write speeds close to SRAM.

Its product line is divided into three parts

Toggle MRAM (older generation, for industrial and military use), STT-MRAM (self-rotating torque, new generation, for data centers), and TMR magnetic sensors. Q1 2026 total revenue was $14.9 million, with MRAM product sales of $14.1 million, up 28% year-on-year, and the remaining $800,000 coming from licensing, royalties, and foundry revenue.

The irreplaceability of this company does not come from size, but from "customers can't leave."

For scenarios like industrial controllers, satellite electronics, and military avionics, the certification cycle for memory typically takes 6-18 months, while in aerospace/defense it can take 3-7 years. Once design is implemented, it is extremely difficult to replace. Combined with ITAR export controls and the "Buy American Act" mandatory requirements for domestic production capacity in the US, Everspin's Globalfoundries foundry and the newly signed second production line for Microchip's Oregon Fab 4 have effectively locked a 5-7 year moat.

So here's the question

Is this moat strong enough to hold up when giants like Samsung and SK Hynix enter the market? Frankly, it's not enough.

After a candlestick, how do you cash out a P/S/11.5x price?

Dividing today's market cap of $633M by FY2025's full-year revenue of $55.2M yields a price-to-sales (P/S) ratio close to 11.5 times—this already marks the valuation ceiling for small-cap semiconductor stocks.

But the context of comparison is as follows

The median P/S ratio for mature semiconductors is about 3-5 times, and foundries enjoy platform premiums at about 8-10 times. At Everspin's current level, the market assumes it is a "foundry platform with a proven business model," not a small company losing 6 million yuan annually and barely scraping the book on interest income.

To be more straightforward

On April 30, Needham raised its target price from $14 to $18.50, with a consensus average target of $18.00. This means that even the most bullish analysts still forecast implied downside after the contract announcement at -31% to -33%.

The story on the cash flow side is equally paradoxical. FY2025 free cash flow of $10 million, which looks good, but net profit for the same year is -$600,000; Q1 2026 operating cash flow plunged to $500,000, down more than $2 million quarter-on-quarter. Such volatility is hard to consider as "stable self-sustaining ability" when valuing.

So what kind of future pricing will the market actually have?—— largely price-in a perfectly executed version of 2027-2028 for the $40 million defense contract + Microchip's ten-year contract contract—two orders that haven't actually materialized.

+154% from the 200-day moving average—is this a breakout or an overshoot?

According to the report, Everspin's fundamentals remain pre_breakout—annualized revenue approaching $60 million, operations not yet breaking even, and key contracts awaiting execution.

But secondary market pricing has already rushed into the breakout phase: the stock price deviated +154% from the 200-day moving average of $10.63, +141% from the 50-day moving average of $11.18, and today hit a 52-week high of $27.63 intraday, and also broke through the company's all-time high of $24.88 set in July 2017—the first time in nine years that MRAM has reached such a high level.

Historical experience tells us that a deviation of more than 50% from the 200-day moving average for small-cap stocks is considered extremely overbought, and a +150% deviation statistically means the median drawdown over the next 12 months is often above 40%. The gap between fundamentals and sentiment has been stretched to this extent, meaning either fundamentals are rapidly catching up or sentiment is returning first—the historical probability of the latter is significantly higher.

The "side business" of big companies and the "everything" of small companies

In the MRAM track, competition is structured in two layers: "big companies' side businesses + professional pure products."

The first layer consists of Samsung, SK Hynix, Micron, and Intel—their eMRAM businesses are just scraps in the embedded memory portfolio, with DRAM, NAND, and HBM as the main battlegrounds. Samsung had already developed eMRAM at the 28nm FDSOI node, but it accounted for less than 0.1% of consolidated revenue, making it a strategic priority so low it could almost be ignored.

The second layer is TSMC, a foundry giant—if it ever decides to bring MTJ modules to its own foundry platform, Everspin's foundry services will be completely taken away. But TSMC's current core pace is 3nm/5nm combined with CoWoS packaging, while MRAM foundry is a "much later" edge project.

So, just how deep is Everspin's moat?

In other words

Its advantage lies in "focus" rather than "scale"—investing 100% of its energy in a niche market that only generates 0.1% of revenue for large companies, in exchange for deep ties with aerospace, defense, and industrial automation clients. This niche is extremely difficult to disrupt within 1-3 years, but it also means it will never become the next Micron.

What is even more concerning is the horizontal replacement of MRAM by new types of memory—once ReRAM and PCM leap over the cost curve, part of the MRAM market in industrial control scenarios will quietly be taken away.

In the pool of $300 million to $800 million, 7%-18% of the share is already the top

The global non-volatile memory (NVM) market size is about $30-40 billion annually, with MRAM as a segment accounting for 1-2%, corresponding to the actual addressable market of about $300-800 million. Everspin's FY2025 revenue is $55.2 million, accounting for a 7%-18% market share, making it a leading independent MRAM vendor.

By endpoint, industrial automation SAM is about $25-40 million, data center persistent memory is about $10-20 million (this segment heavily relies on AI narratives), defense/space radiation hardening is about $20-80 million (Everspin's truly highly sticky market), and automotive and medical combined are in the $15-25 million range.

The outlook three years ahead should be viewed by scenario: under optimistic conditions, Microchip foundry + Amentum contracts + defense orders will simultaneously increase volume, with FY2027 revenue expected to reach $75 million-95 million, with a CAGR of 11%-18%; If Catalyst's delivery pace falls short of expectations (the probability is not low), revenue may stagnate between $60 million and $65 million, with growth dropping to 5%-8%.

Three-hit narrative: Microchip, Amentum, xSPI mass production

In the next 12-24 months, Everspin's stock price will mainly be driven by the following events: whoever cashes first will be rewarded by the market, while whoever delays will be punished:

• $40 million Amentum subcontract execution: Starting April 20, 2026, ending November 21, 2028, payments will be made in two milestone phases. The impact is positive, but the amount has basically been price-in
• Microchip Oregon Fab 4 Mass Production: The first batch of products is expected to roll off the line in the second half of 2027, marking a key milestone for the company's upgrade from a single foundry partner to a "dual supply chain."
• High-reliability xSPI 128Mb / 256Mb mass production: Certification and mass supply to be completed in the second half of 2026, directly connecting industrial and aerospace customers
• $14.6 million DoD Maintains Contract End: Expected to end in the first half of 2027, which is a downside risk point for revenue base
• Q2 2026 Financial Report: The company provides revenue guidance of $15.5M-$16.5M, Non-GAAP EPS of $0-$0.03, expected to be released on August 6
• Customer concentration disclosure: If the proportion of Customer A/B accounts receivable decreases, it indicates an improvement in customer structure; If the level remains high, concentration risk will persist

It took 9 years to return to $24—can it hold steady this time?

Looking at the longer term, since Everspin was listed at $8.02 in October 2016, its stock price has mostly been stuck in the $5-$15 range, trading sideways with low liquidity and minimal institutional attention. On July 17, 2017, the stock closed at a historic high of $24.88, but it took nearly nine years to reach that level again.

The real turning point will come from 2024 to 2025. @PepInvestStocks and other deep value accounts have begun to incorporate MRAM into the AGI infrastructure narrative. Aschenbrenner's "Situational Awareness" report lists persistent memory as a key track for non-GPU investments, combined with the $14.6 million DoD contract maintenance on August 14, 2024, and the Microchip Oregon 10 announced on April 8, 2026 The Centroid agreement and the $40 million Amentum subcontract signed on April 24—three major events have been intensively fulfilled, with the stock price surging from $5.49 to $27.63 over the past 12 months, an increase of over +400%.

But another lesson history offers is equally worth being wary of

Everspin was founded in 2008 and took 16 years to reach annual revenue exceeding $50 million. The nonlinear pattern of new semiconductor technologies from "concept" to "volume ramp-up" means that every market "turning point narrative" often occurs 3-5 years before actual realization. Whether this time is truly different still requires the full-year FY2026 financial report and actual data from Microchip's production line to verify.

⚠️ Risk Notice First, valuations have already reached the entry level of a bull market scenario. The implied growth rate of 11.5x P/S far exceeds the FY2025 actual +9.5% fundamental growth. Any contract extensions, orders falling short of expectations, or competitors breaking through new memory systems could trigger a double blow of "valuation compression + underperformance performance." Second, on May 4, 2026, CEO Sanjeev Aggarwal sold 28,459 shares at an average price of $19.58, and on May 7, CFO William Cooper sold 11,000 shares at nearly $21.75. The core management's systematic sell-off near record highs signals significant signs; Third, the overlap of customer concentration risk and small-cap liquidity risk—with only about 23.55 million shares outstanding, any design elimination or procurement adjustment by a major client would have an immediate impact on revenue, and the ongoing $1.6 million litigation costs in Q1 will continue to erode operating profit for at least the next two quarters. The content described in this article is solely based on research and analysis based on publicly available information and does not constitute any investment advice. Investing involves risks; please proceed cautiously. All decisions should be made independently based on personal research and risk tolerance.

💬 Discussion CEOs are all selling—are you willing to keep buying?

Data source

• Latest stock price, 52-week range, intraday highs and lows: Robinhood MRAM Market Page https://robinhood.com/us/en/stocks/MRAM/
• Daily Turnover, Market Cap, Beta: TradingView NASDAQ: MRAM https://www.tradingview.com/symbols/NASDAQ-MRAM/
• Analyst target prices (average $18.00, Needham $18.50): Investing.com / MarketBeat https://www.investing.com/equities/everspin-technologies-inc
• Q1 2026 Financial Report Data (Revenue, Gross Margin, EPS, Cash): SEC 8-K Original https://www.sec.gov/Archives/edgar/data/0001438423/000162828026028376/mram-20260331xexx991.htm
• $40 million Amentum subcontract terms: SEC 8-K original https://www.sec.gov/Archives/edgar/data/0001438423/000162828026028432/mram-20260424.htm
• Microchip's 10-Year Oregon Fab 4 Foundry Agreement: BusinessWire Original https://www.businesswire.com/news/home/20260430725216/en/Everspin-Executes-$40M-Agreement-for-Mil-Aero-MRAM-Applications
• CEO/CFO share reduction disclosure: Combined https://www.tickerreport.com/banking-finance/13433806/everspin-tech-nasdaqmram-reaches-new-12-month-high-heres-why.html from TickerReport / GuruFocus
• Q1 2026 earnings call highlights (litigation costs, production line schedule): Yahoo Finance / The Motley Fool https://finance.yahoo.com/markets/stocks/articles/everspin-technologies-inc-q1-2026-123000886.html
• All-time closing price $24.88 (2017-07-17): Macrotrends MRAM's 10-year stock price https://www.macrotrends.net/stocks/charts/MRAM/everspin-tech/stock-price-history