Product Growth Delivers; Cash Costs Temper the Defense Upside

Results vs. Consensus

Q2 2026 Scorecard

Year-over-Year Comparison

Quarter-over-Quarter Comparison

Revenue Assessment

Q1’s upgrade depended on growing product shipments and turning the $40M defense agreement into revenue. Both conditions were met. Product revenue increased $1.213M sequentially, while nonproduct revenue climbed to $3.423M from $0.772M. The reported beat over the $16.5M guidance ceiling is meaningful, although the original guide excluded the new subcontract. The stronger product line is the cleaner evidence of underlying customer demand.

Margin Assessment

The conclusion that product economics had permanently moved above 50% no longer holds. Product gross profit fell despite rising sales because its margin declined about 485 basis points. Nonproduct revenue carries a roughly 90% gross margin this quarter and grew to 18.3% of sales from 5.2%, lifting the consolidated average. We expect company margins to remain above 50%, supported by service mix, while valuing the product business on a more cautious high-40s margin assumption.

EPS and Cash Assessment

Adjusted operating income rose to $2.068M from $0.213M in Q1, a substantial improvement before excluded expenses. Interest and other income fell $1.629M as the older defense sustainment contract wound down, leaving adjusted net income up only $0.229M. The reported $0.11 EPS also reflects approximately 25.9M diluted shares; the 23.9M GAAP loss denominator excludes antidilutive awards and is unsuitable for valuing positive adjusted earnings.

The new engineering-fee exclusion adds another cost of expansion outside adjusted earnings. Retaining that fee would leave approximately $1.812M of adjusted net income, or $0.07 per diluted share, still above the prior $0.03 guidance ceiling. These fees and litigation consume resources even when they are excluded from management’s performance measure. Our forecast therefore retains separate GAAP and adjusted outcomes, and our valuation assumes less cash than the June balance.

Segment Performance

Everspin manages one reportable segment. Its product and nonproduct revenue categories reveal the economic mix; end markets are described qualitatively.

Products: Recovery Broadens Beyond the Q1 Leaders

Industrial automation in Japan and energy management in Europe drove the recovery, alongside broad aerospace and defense demand. These are applications where persistence through power loss and frequent writes can justify MRAM’s cost premium. Product growth accelerated from 27.9% in Q1 to 38.1% in Q2, supporting the earlier design-win conversion thesis without requiring a new mass-market application.

Assessment: We raise FY26 product revenue to $63.5M. The next test is shipment growth sufficient to cover higher backend costs: at the Q3 revenue midpoint and roughly flat nonproduct sales, products need to reach about $16.6M, up 8% sequentially.

Nonproduct: Contract Recognition Changes the Mix

The new defense subcontract generated a substantial majority of nonproduct revenue, rather than an independently disclosed $3.423M contract contribution. Its engineering work produces high reported gross profit but comes with a project schedule. This supports several quarters of earnings visibility; it does not establish a permanent royalty stream or a guaranteed quarterly minimum.

Assessment: We expect about $3.4M of nonproduct revenue in Q3 and $3.5M in Q4, below the earlier assumption of $4M per quarter from the subcontract alone. The lower service assumption is more than offset by the stronger product outlook.

End-Market Read-Through

Key KPIs

Key Topics & Management Commentary

Overall Management Tone: Management remained confident about demand and delivery, consistent with Q1. The more guarded answers concerned revenue timing, product costs and manufacturing continuity, making the tone on returns less assured than the tone on growth.

1. Defense Execution Improves; Follow-On Awards Remain Optional

Beginning revenue recognition fulfills the first operational commitment behind Q1’s upgrade. The older $14.6M sustainment contract is separately winding down, with just $0.5M recognized in other income during Q2 and completion still expected in H1 2027. The new contract must replace that lost below-the-line support before its contribution becomes entirely incremental to net income.

“So as you know, we've had a few contracts now over the last 5 years with the U.S. government. So we work very closely with them, keeping them informed of the technology development that we're doing at Everspin. And whenever there is an overlap between the goals of the U.S. government and Everspin's road map, it typically leads to a first a small contract and then a bigger contract to actually do the development. So we do have a few irons in the fire, but there's nothing concrete yet.”
— Sanjeev Aggarwal, CEO

Assessment: The domestic manufacturing relationship is valuable, but another large award is upside to our forecast. The existing subcontract supports our base case; it does not justify assuming a sequence of new contracts at the same margin.

2. Litigation Is a Larger Cost Than Q1 Allowed

Quarterly litigation expense increased by $2.398M, or 147%, and management includes a similar roughly $4M in Q3. Q1’s description of a bounded $1.6M run rate proved too optimistic. Excluding legal expenses explains much of the adjusted profit, but offers no relief to cash funding needs.

Assessment: We now assume roughly $13.7M of FY26 litigation costs, including another $4M in Q4 as an analyst assumption. The FY27 earnings recovery depends partly on these costs easing; a reduction is a forecast sensitivity, not an announced resolution.

3. PERSYST Delivers Ahead of Schedule

The higher-density PERSYST rollout improves on the H2 availability target carried from Q1 and gives customers a broader range without waiting for UNISYST.

“During the second quarter, we released 128-megabit high reliability parts to production. Subsequent to the quarter end, we released all SKUs of xSPI 256 megabit density to production, including high reliability parts. Customers now have these parts on hand to evaluate them in their designs.”
— Sanjeev Aggarwal, CEO

Assessment: Product availability reduces roadmap risk and supports FY27 product growth. Customer evaluation still separates a released part from volume revenue, so we avoid assigning unsupported sales to individual densities.

4. UNISYST Samples Move Into 2027

The 256Mb xSPI test chip remains scheduled to tape out later this year on a 16nm process. Engineering samples are now expected in early 2027, compared with Q4 2026 in the prior call, with production later in 2027. The shift matters because customer qualification follows sample delivery; commercial availability is only the start of the adoption cycle.

Assessment: We move the UNISYST timing pillar to at risk and exclude meaningful contribution from FY27. The $3B addressable-market ambition remains a long-term opportunity, with no near-term earnings attached to the proposed 5–10% share.

5. The Second Fab Becomes More Important

Microchip work began in April, with equipment installation and process-gap analysis underway. Management targets first qualified silicon 18–24 months after kickoff, placing that milestone around October 2027–April 2028. This is a broader window and an earlier technical milestone than the prior investment case’s H2 2027 revenue start.

The proposed ownership change at Chandler adds urgency. Management expects operations to continue through the end of 2028 under NXP ownership and plans discussions with Nokia about the period thereafter. A second site can protect continuity and expand capacity, but only after qualification.

Assessment: The manufacturing investment is more strategically necessary, while its direct revenue timing is less certain. We treat meaningful Microchip output as a 2028 contribution and flag post-2028 Chandler access as a new risk within the manufacturing pillar.

6. CXL Is an Engineering Milestone Before It Is a Sales Story

Everspin has a controller-IP development contract and is building an FPGA demonstration using AMD’s UltraScale+ platform. Its MaxLinear memorandum covers evaluation of persistent memory for metadata, logs, buffers and caches. Those tasks can be bottlenecks despite representing a small share of stored data, making low-latency persistence a plausible source of system value.

Assessment: September’s planned demonstration should test protocol operation and system performance. We assign no FY26–27 revenue to CXL: efficiency targets and an evaluation agreement support technical exploration, while a deployable design, customer qualification and paid volume orders determine the financial opportunity.

7. The Cash Increase Came From Equity Plans

Cash increased despite negative free cash flow because option exercises and employee purchases supplied $5.079M. In H1, receivables and inventory together consumed $5.021M of cash, partly offset by deferred revenue and accrued liabilities. The balance sheet has capacity to fund expansion, but this quarter does not show that operations can yet fund it unaided.

Assessment: We expect roughly $38M of year-end cash, allowing for another $5.7M of H2 cash use before financing. Our valuation uses $35M of forward net cash and 26M diluted shares, recognizing both investment requirements and the claims of equity awards.

Guidance & Outlook

Q3’s midpoint implies 6.7% sequential sales growth. With nonproduct revenue expected near Q2, about $1.3M of that increase must come from products. The result would extend shipment momentum even as the initial step in service revenue levels off.

Our outlook: We expect $20M in Q3 and $21M in Q4, taking full-year revenue to $74.6M. The implied H2 average of $20.5M is achievable if the product recovery continues and service milestones remain near their current pace. Consolidated GAAP gross margin around 54% in Q3 comes from product margin near 47% and nonproduct margin near 88%, rather than a return to Q1 product economics.

Guidance style: Q2’s strong beat benefited from a contract explicitly absent from the initial guide. That makes a similarly large Q3 beat a poor default assumption. A new delay in service milestones would expose the product business’s thinner margin cushion.

Analyst Q&A Highlights

Contract Revenue Will Not Simply Scale With a Full Quarter

Q: “And then the second one I wanted to ask was about the $40 million contract. So last quarter, you only had about, if I remember correctly, 2 months of that was recognized in the quarter.”
— Neil Young, Needham & Company

A: “That's right.”
— Bill Cooper, CFO

Q: “So should we expect maybe another step-up now in 3Q that's a full quarter? Or is it not going to scale sort of in an evenly manner?”
— Neil Young, Needham & Company

A: “Yes, that's correct. It won't necessarily scale in a very linear manner. So I would expect to see nonproduct in the similar area from Q2 to Q3.”
— Bill Cooper, CFO

Assessment: A full quarter of activity does not automatically produce more recognized revenue. This directly challenges the earlier $4M quarterly subcontract assumption and makes product shipments the main source of near-term sequential growth.

Product Costs Can Offset Manufacturing Improvements

Q: “As I oftentimes ask here, I noticed in the second quarter, your product gross margins were a bit lower than the first quarter and kind of similar to the range you had in most of 2025. I want to get a sense of kind of the forward outlook there. Is this kind of the baseline to think? Or can we get back towards that 50% level? Just kind of high level, how do you think about that?”
— Richard Shannon, Craig-Hallum

A: “And so what we've always guided is, hey, we expect product gross margins to kind of be in that mid to upper 40s range. And then in total, we expect the company to be north of 50% for total gross margins.”
— Bill Cooper, CFO

Assessment: Management answered with distinct product and company margin ranges. The CEO then explained that packaging and testing price increases absorbed internal improvements, and affirmed the pressure could persist. This points to an input-cost problem beyond temporary low utilization. Our high-40s product-margin forecast consequently requires some recovery without assuming full reversal to Q1.

The Next Quarter Still Includes Heavy Legal Spending

Q: “So the difference here between the pro forma and the GAAP EPS here, I'm assuming this is mostly from legal expenses. I know you quantified this roughly $4 million in the second quarter. I didn't have time to do the math here, but is that a similar number that's baked into the third quarter as well? Or how do you think about that?”
— Richard Shannon, Craig-Hallum

A: “Yes. Yes, that's correct. We baked in a similar number.”
— Bill Cooper, CFO

Assessment: The concise answer is a concrete Q3 expense commitment. It does not supply an end date. We retain the expense through Q4 in our forecast, preventing adjusted EPS growth from being mistaken for near-term cash liberation.

Chandler Continuity Extends Through 2028, With Later Access Unsettled

Q: “Sanjeev, I noticed that NXP has sold the -- or has an agreement to sell the Chandler fab. And obviously, noting that you've already have an agreement with Microchip to expand capacity here. How do you think about this in the context of your needs here? Can you -- when the Chandler fab conveys over completely, do you expect to be out of there or not? And to what degree does Microchip alone or do you expect them to be able to cover your needs for the products that are affected -- possibly affected by the Chandler fab sale?”
— Richard Shannon, Craig-Hallum

A: “So we don't see any interruption to our operations, at least through the end of 2028. And we are in conversations or we have some planned conversations with Nokia to understand what are their plans for Everspin. We have heard positive things, but we haven't directly spoken to them yet. So in an ideal case scenario, we would have both facilities. And if the business requires us to keep both facilities, that would be great. And if not, then we obviously brought on Microchip so that we can actually scale production if Nokia had other plans for the fab.”
— Sanjeev Aggarwal, CEO

Assessment: Management distinguishes continuity under the expected ownership timetable from future access it has not negotiated. Microchip offers an alternative if Nokia’s plans conflict with Everspin’s needs. The answer supports near-term supply but raises the importance of completing the second-site qualification before that protection expires.

A GEO Win Does Not Establish Universal Radiation Qualification

Q: “I didn't get a chance to ask you about this after the announcement with Astro, I forget their full name with the win for -- this is for GEO satellites. I think this is your first win in the GEO area after having talked about LEO satellites a lot. Let me get a sense of the importance of that win. And ultimately, do you see the opportunity here being bigger for GEO, LEO, MEO or just kind of characterize the opportunity holistically in satellites, please?”
— Richard Shannon, Craig-Hallum

A: “This is our first design win for a GEO satellite mission using our commercially developed MRAM. I mean, obviously, it's qualified for extended temperatures. But we do -- we have not done any radiation hardening for these parts that Astro Digital has designed in their satellite mission. […] So there must be some redundancy or I don't really know what they're doing.”
— Sanjeev Aggarwal, CEO

Assessment: The answer establishes an additional use case but does not rank the GEO, LEO and MEO markets as asked. Customer-specific system design may be essential; the win is not evidence that every satellite platform can adopt the same part unchanged.

CXL Revenue Depends on a Working Prototype and Identified Design

Q: “And then I guess just lastly, as far as the CXL interface and 3-year product road map you mentioned, can you just provide some color on what that rollout might look like and kind of the external guide points we might see?”
— Josh Sullivan, JonesTrading

A: “So I think it's a huge market, but it's a little bit early for me to say how the revenue will build up over the next 3 years or so. So I think once we have the prototypes working and we have a design identified, I think then we can talk about projections of revenue and percent of market capture.”
— Sanjeev Aggarwal, CEO

Assessment: The substantive answer places prototype operation and design selection ahead of revenue forecasts. Management’s hoped-for accelerator-efficiency improvement remains a target for testing. That preserves significant upside while supporting our decision to base valuation on established products and contracted services.

What They’re NOT Saying

1. A subcontract revenue and margin schedule: Neither remaining milestones nor a life-of-contract margin is quantified. A quarter near 90% nonproduct gross margin cannot establish the economics of all remaining engineering work.
2. A litigation resolution date: Q3 spending is addressed, but an exit timetable is absent. Continued costs at $4M per quarter would materially reduce the earnings available to shareholders.
3. Measured CXL system gains or paid customer orders: The prototype work targets meaningful efficiency improvements; the call does not provide validated results or committed production demand.
4. Long-term Chandler access: Management has positive indications, but no settled post-2028 arrangement with the prospective owner.
5. Quantified NOR conversion or rail follow-through: The earlier use-case thesis survives; the call does not establish its incremental revenue contribution or erase the MRAM cost premium.

Market Reaction

• Pre-print setup: MRAM closed August 5 at $16.14, up 73.9% year to date and 170.8% over 12 months, but down 17.7% over 30 days. Its preceding 52-week closing range was $5.94–44.01.
• Reaction session, August 6: The stock opened at $15.58, traded between $15.05 and $16.61, and closed at $15.65, down 3.0%.
• Participation and benchmark: Volume was 1.8M shares versus a 1.4M 30-day average, about 1.3 times normal. The S&P 500 declined 0.2%.

The negative response came despite a revenue beat and another sequentially higher guide. In our view, it is consistent with investors demanding better cash economics after a large year-to-date appreciation. Litigation and weaker product margins offer a plausible reason for caution, but the price move alone cannot identify which concern drove selling.

Contemporaneous bullish commentary emphasized CXL development and space traction, with one broker raising its target modestly while retaining its positive rating. That optimism concerns the opportunity set; Q2’s financial statements place a lower bound on how quickly the opportunity can become cash. The pullback improves prospective returns without resolving that tension.

Street Perspective

Debate: Does Defense Revenue Deserve a Durable Premium?

Bull view: Initial service recognition and accelerating shipments strengthen the case that Everspin’s defense relationships can generate a broader, more valuable franchise.

Bear view: Project revenue is irregular, the older contract’s income is fading, and follow-on awards remain uncertain. A temporary service mix boost deserves a lower multiple than recurring royalties.

Our take: Existing commitments support a premium over a purely cyclical component supplier, but we exclude unawarded contracts. A modest premium is justified by execution already visible; a compounding series of awards is not yet earned.

Debate: Can Adjusted Profit Become Cash Profit?

Bull view: Adjusted operating margin reached 11.0%, showing leverage as sales absorb core spending. Legal and initial engineering costs could diminish once expansion matures.

Bear view: Those costs are required or recurring over the investment horizon, and equity-plan proceeds are funding the cash balance. Excluding them overstates current shareholder earnings.

Our take: Operating leverage is real, but cash conversion deserves more weight than the adjusted EPS beat. We expect a GAAP loss in FY26 and only modest GAAP profitability in FY27, with substantial sensitivity to legal spending.

Debate: How Much AI and Space Optionality Belongs in Today’s Price?

Bull view: CXL controller development, the MaxLinear evaluation and a GEO design win widen the set of commercially relevant applications.

Bear view: Prototypes, evaluations and individual design-ins do not establish large production ramps. UNISYST’s later sample timing illustrates the distance between a product concept and revenue.

Our take: These projects strengthen long-term relevance but remain outside our near-term sales forecast. A working CXL demonstration followed by an identified customer design would reduce risk; a performance target alone does not support a higher revenue estimate.

Our Estimates and Valuation

The Earnings Bridge

FY26 revenue comprises $33.608M already reported, $20M in Q3 and $21M in Q4. Our H2 product/nonproduct splits are $16.6M/$3.4M and $17.5M/$3.5M. With product gross margin of 47%, nonproduct margin of 88%, modest cost-of-sales stock-compensation addbacks, and adjusted operating expenses of $8.35M/$8.60M, H2 adjusted operating income is about $5.47M. Adding $1.14M of interest and other income to H1’s $5.495M adjusted net income gives $12.10M for the year, or approximately $0.47 per diluted share.

Deducting approximately $13.68M of legal costs, $5.47M of stock compensation and $1.55M of engineering-fee exclusions gives an $8.60M GAAP loss, or about $0.35 per share. The loss denominator is lower because antidilutive awards are excluded. Q3 adjusted EPS is roughly $0.125; GAAP EPS near $(0.10) is at the cautious end of management’s range.

The $72M product assumption represents 13% growth from FY26, substantially below the current quarterly pace. The $14M service estimate does not require a new award. FY27 adjusted OpEx of $34M rises from our FY26 estimate of $32.91M but is nearly flat against the $34.4M annualized Q4 run rate. We assume the core team and development capacity can absorb 13% product growth; faster investment would reduce that leverage. Each additional $1M of expense lowers diluted EPS by about $0.04. Lower litigation spending is a separate assumption supporting GAAP profitability. If annual legal costs stay near $16M instead of $6M, roughly $10M of earnings and cash disappear before other effects. A one-point change in FY27 product margin changes annual pretax earnings by $0.72M, or about $0.03 per diluted share.

A Lower Valuation Despite Higher FY26 Sales

We reduce 12-month fair value to $18–22 from $24–30. The earlier range gave more credit to product-margin improvement, rapid service growth and near-term manufacturing contribution than this quarter supports. The larger prospective diluted share count also reduces value per share. Our current framework uses enterprise value to revenue, with an earnings cross-check, and assigns no separate value to uncommercialized CXL or NOR-market share goals.

Applying 5.0–6.2 times FY27 revenue of $86M, adding $35M of forward net cash and dividing by 26M diluted shares produces $17.9–21.9 per share. These are our valuation assumptions. The premium reflects double-digit product growth, contracted service work and a potentially valuable domestic supply position; the upper end requires evidence that adjusted operating leverage can outlast the investment phase. At about $20, the stock would still trade near 35 times our adjusted FY27 EPS, so successful cost normalization is already part of fair value.

Using analyst scenario weights of 25% downside, 50% base and 25% upside gives approximately $19.3 of expected value, or a 23% return. We assume no dividend. That exceeds our illustrative 8% S&P 500 total-return hurdle over the same 12 months and supports Outperform at $15.65, with moderate 6/10 conviction. The roughly one-third downside is credible: weaker mix, continuing litigation and further dilution could combine with multiple compression.

Thesis Scorecard Post-Earnings

The standing case was last expressed in our Q1 recap. The same adoption, execution and risk questions carry forward below; the current status reflects Q2 evidence.

Overall: Revenue adoption is stronger, while the margin, cost and timing assumptions supporting Q1’s valuation have weakened. Q2 exceeded the earlier $16.5M sales signpost and began subcontract recognition. It did not meet the additional-follow-on-award condition, and the earlier description of litigation as bounded is no longer tenable.

What changes the view: Q3 delivery within $19.5–20.5M with product-led growth and nonproduct near $3.4M would reinforce our forecast. Legal spending materially above the roughly $4M embedded in guidance, a subcontract milestone delay, or further slippage from early-2027 UNISYST samples would reduce it. Lower legal costs combined with operating cash funding capital investment would justify restoring valuation confidence; CXL prototype success alone would not.

Action: Maintain Outperform, with lower conviction and $18–22 fair value over 12 months. At $15.65, established product growth and signed service work offer enough prospective return to compensate for the risks, provided the investor can tolerate the approximately $10 downside case. The investment case now rests on converting growth into cash, as well as delivering growth itself.