Silicon Motion Sells the Boot Drive It Swore It Wouldn't Sell

A record quarter, a 40%-plus 2027 guide, and a quiet reversal into the DRAM-based server boot drive market management once ceded to the NAND makers

SIMO · Earnings Call · 2026-10-29

The quarter that reset the base rate

Silicon Motion's September quarter was not a normal beat. Revenue came in at $560 million, up 24% sequentially and 131% year-on-year, well above a $519–541 million guide. Gross margin of 53.7% blew through the outlook, operating margin reached 32.1%, and non-GAAP EPS of $4.35 per ADS rose 79% sequentially and 333% year-on-year. Jason Tsai did not soften the framing: “we believe the momentum in our business will continue in the fourth quarter and now expect revenue to grow 5%-10% sequentially to $588 million-$616 million” — Jason Tsai, Chief Financial Officer (CFO) · 2026-10-29, with gross margin guided up again to 54–55% and operating margin to 36–37%. Then came the number that matters more than any single quarter. Management guided 2027 to “grow revenue by at least 40% next year, with gross profitability to be in the low to mid 50% range” — Jason Tsai, Chief Financial Officer (CFO) · 2026-10-29. A company that spent years telling investors to model a 48–50% gross margin ceiling just guided to a structurally higher one. On the July call, Tsai was explicit: “we're still comfortable with the 48% to 50%... long term, we still expect to be in that 48% to 50%” — Jason Tsai, Chief Financial Officer (CFO) · 2026-07-30. Three months later, that framework is gone, replaced by a mix story: solutions — Ferri automotive plus enterprise boot drives — reached over 40% of revenue in Q3 and should approach 50/50 with controllers in 2027.

The reversal nobody flagged

The most interesting thing on this call was not the guide. It was a market the company spent the last several quarters explicitly declining to enter. In July, Wallace Kou drew a bright line around onboard DRAM boot drives: “near CPU Boot Drive with conventional enterprise controller with DRAM together... That business belongs to NAND maker. It's conventional, traditional. We don't compete that sector.” — Chia-Chang Kou, President and Chief Executive Officer (CEO) · 2026-07-30 The whole CPU boot thesis was built on the DRAMless boot drive niche that NAND makers had abandoned as too small and too fiddly. On this call, that line vanished. Kou now argues the NAND makers have vacated the DRAM-based near-CPU socket too, because every wafer is being pulled toward HBM and 16-terabyte compute drives.

That will be a minimum $10 billion by 2030, because the density is bigger than the DRAMless boot drive... there is lack of resource and attention in the near CPU boot drive with DRAM. So that's why the supply becomes shortage, and many tier one customer come to Silicon Motion.
Wallace Kou, President and Chief Executive Officer (CEO) · 2026-10-29

The mechanics are plausible — the SM8000A already carries OCP features, so no new ASIC is required, and ramp is targeted for the second half of 2027. But make no mistake: this is a deliberate expansion from a ~$3 billion 2026 boot-drive TAM heading to over $5 billion by 2030 into a second, larger $10 billion-plus pool, entered from a standing start against incumbent suppliers. The words "we don't compete that sector" have a short half-life. The same logic stretches to Robotics and humanoids, where Kou's throwaway line — “our early conversation suggests robotic opportunity may be larger than automotive” — Wallace Kou, President and Chief Executive Officer (CEO) · 2026-10-29 — is the kind of claim that deserves a skeptical raised eyebrow given the segment's revenue today is, in his own words, "very small."

The engine everyone is ignoring: NAND scarcity

The uncomfortable truth underneath the growth is that Silicon Motion is currently a beneficiary of a shortage it does not control. The NAND pricing backdrop is what pushed solutions revenue up over 75% sequentially — these are controller-plus-NAND bundles, so rising memory prices flow through the top line. It is also what is now breaking the oldest part of the business: eMMC and UFS, which grew over 100% year-on-year in the first half, declined sequentially and are guided roughly flat, as smartphone OEMs despec devices to cope with memory and storage costs. That is a genuine deceleration hidden under a record headline, and it is a repeat of the seasonal-mix problem the company has flagged before — in April, Tsai noted mobile controllers "tend to be a little bit below corporate average." The offset is that Enterprise SSD demand, and the KV cache build-out for inference, is where the incremental dollar now lives. SiFive-style claims aside, the read-through from peers in the same reporting window — SanDisk calling out KV cache and AI inference architecture, memory pricing across the group — suggests this is a sector-wide wave, not a SIMO-only insight. Silicon Motion is riding it with a differentiated position, not creating it. What is company-unique is the balance sheet. The $1.15 billion convertible bond — 0% coupon, 65% conversion premium at $380.50 — lifted cash from $181.8 million to $953.6 million. Tsai says it funds working capital for a couple of years, against a two-quarter inventory turn. Investors should watch that covenant of confidence closely: inventory has been built aggressively, and a two-quarter cash cycle in a market where channel partners are paying peak prices is exactly where a growth story can go wrong.

The tape disagrees

The market context is the sharpest contrast in this file. Across the 90-day window, the memory-and-interconnect complex — high bandwidth memory, high-performance compute, co-packaged optics, data center capacity — is deeply negative on the global tape, with Silicon Motion's own name appearing inside several decliner baskets. Yet the fundamentals here are accelerating, not softening. Either the tape is wrong about SIMO, or the tape is pricing an inventory cycle that has not yet shown up in a transcript. This is the central tension: record results and a raised multi-year margin framework, against a stock whose peers have already de-rated and a company that has publicly admitted it cannot forecast the NAND supply picture beyond a few quarters.

The bottom line

Silicon Motion has genuinely transformed its mix — automotive certifications that take years to build, enterprise controller design wins that carry three-to-five-year product cycles, and now an opportunistic push into a boot-drive socket it formally disclaimed. The margin reset looks structural. But the growth rate is being amplified by a shortage, the mobile business is contracting, the Singapore arbitration has been awaiting a ruling since March closing arguments, and the tape is voting no. The company's own conservative DNA shows in the hedge: "at least 40%" and "low to mid" are not commitments, they are floors. That is the right posture for a business that has just discovered its second act depends on someone else's allocation decisions.