Silicon Motion (SIMO) Q3 2026 earnings review

Silicon Motion's sales rise 131%, but its inventory climbs faster

Silicon Motion's storage parts for cars and AI servers are selling faster than it can collect the money. Sales rose 131% from a year ago, and drives the company builds with bought-in flash memory passed 40% of sales, chief executive Wallace Kou said. Operating profit set a record, yet cash still left the business.

At a glance
Revenue$560.2 million +131% from a year ago
Profit per share, excluding one-offs$4.35 $1.00 a year ago
Operating margin, excluding one-offs32.1% 23.1% last quarter
Fourth-quarter sales outlook$588โ€“616 million middle of the range: 7% growth from this quarter

โš–๏ธ Verdict: ๐ŸŸข Bullish

The case got better, because management's own plan moved up sharply: finance chief Jason Tsai now sees an operating margin above 36% next quarter, where in July he spoke only of clearing 30% by year end. One caution: the cash drain widened, and unsold memory has never been larger against sales.

The question now is whether this is a new kind of company or a well-timed trade in scarce memory. One side says flash makers have left the low-capacity markets for good; the other says today's prices do the work and reverse when supply returns. Fourth-quarter cash flow will show which.

๐Ÿ‚ Bull Case

๐ŸŸข๐ŸŸข strengthening GROWTH

Car and Server Storage Passes 40% of Sales

Silicon Motion sells two things: controller chips, and complete storage drives it assembles by pairing its own controller with flash memory it buys in. The drive side covers Ferri, built for cars, drones and robots, and boot drives, the small drives that start up servers and networking equipment.

  • Growth from the previous quarter: 75% to 80%
  • Growth from a year ago: 2,580% to 2,585%
  • Share of sales: over 40%, against about 30% three months earlier
  • SSD controllers, the other engine: up 20% to 25% from the previous quarter

This matters because a drive carries the price of the memory inside it, so one design win brings several times the revenue of a controller alone. Kou said the order book has grown significantly in the past three months, and finance chief Jason Tsai described car work as "multi-year engagements where, once we are qualified in, we have got a really long-standing position with that socket".

What to watch: the growth rate of this line in the fourth quarter, which Tsai expects to lead again. Double-digit sequential growth, with cars in front, would show the automotive and server design wins ramping rather than customers stocking up before prices rise.

๐ŸŸข๐ŸŸข strengthening MARGIN

Research Costs Shrink Against Sales

Designing flash controllers is expensive, and that spending used to eat most of Silicon Motion's gross profit. In the September quarter research and development took 20.7% of sales, against 28.7% a year earlier. The dollar amount climbed by two thirds, while sales more than doubled.

  • Cost of making the products: 46.4% of sales, 5.0 points lower than a year ago
  • Research and development: 20.7% of sales, 8.0 points lower
  • All operating costs: 24.6% of sales, 11.9 points lower
  • Operating margin excluding one-offs: 32.1%, against 15.8% a year ago

Mix did the rest. Kou credits car storage and enterprise boot drives, which earn more than the consumer controllers beside them, and Tsai pointed to "better-than-expected margins in our Ferri automotive products". Stock-based pay of $17.3 million, five times June's charge, is the one cost that ran the other way, and it sits outside the margin excluding one-offs.

What to watch: operating costs as a share of sales in the fourth quarter. The company's own plan implies about 18%, by our math, down from 21.5% this quarter. A reading near that would show the leverage is structural, not a gap between chip tape-outs.

๐ŸŸข new GROWTH

Silicon Motion Targets 40% Growth in 2027

For the first time the firm put a figure on next year. Tsai told analysts "we can grow revenue by at least 40% next year, with gross profitability to be in the low to mid 50% range". By our math that is sales above $2.7 billion.

Supply backs the claim, not demand alone. Kou said "we secured our 2027 wafer with TSMC primarily", with test, assembly and substrate capacity too. He expects drives and controllers each near half of sales next year, and the 4-nanometer Gen 6 controller taped out on the date promised in July.

What to watch: the detailed 2027 plan Tsai set for January. If it holds gross margin above this quarter's 53.7% while sales grow 40%, the shift toward cars and servers is adding profit, not only volume.

๐Ÿป Bear Case

๐Ÿ”ด๐Ÿ”ด persistent CASH

Inventory Now Sits at Ten Months of Costs

Because Silicon Motion now sells finished drives, it buys the flash memory first and carries it until a customer takes delivery. That stock reached $870.9 million at the end of September, up from $338.0 million a year earlier. Set against the quarter's cost of sales, it equals 305 days of supply by our math.

  • Days of cost of sales: 305, against 273 three months earlier
  • A year ago: 247 days
  • Cash tied up in this quarter's build: $197.9 million

The risk is price. Kou said flash prices in the third and fourth quarters are "rather similar" and sees moderate increases next year, so this memory was bought near the top of a shortage. Tsai answers with speed: he is "targeting about two quarters turnaround time between when we get the NAND and when we generate the cash flow".

What to watch: inventory days in the fourth quarter, after Tsai said the stock will grow again before it turns. Days above 305 with gross margin holding would mean the firm is buying memory against orders it already holds.

๐Ÿ”ด๐Ÿ”ด persistent MACRO contradicts narrative

Consumer Chips Shrink as Memory Prices Bite

The release treats the shortage as an opening: memory makers are "pulling back from certain consumer-oriented segments", which "creates a meaningful opportunity for our products in these segments". Silicon Motion's own consumer lines read differently.

Controllers for phones and connected devices, the eMMC and UFS chips that manage storage inside a handset, fell 10% to 15% from the previous quarter. Kou blamed launch timing and handset makers struggling to price devices built with costly memory. "We expect the headwind to persist in the near to mid terms," he said.

On the larger line, Tsai said: "We believe our SSD controller business will be flat to slightly down given the current memory and storage environment and typical seasonality." Those controllers grew more than 20% this quarter, so the fourth quarter takes away the engine that worked. Share gains continue, because the phone chips sit 35% to 40% above last year, but direction now follows the end market.

What to watch: the eMMC and UFS line in the fourth quarter, and the UFS 4.1 and UFS 5 chips Kou says restore growth next year. A second decline there would leave the story resting on enterprise orders.

๐Ÿ”ด persistent CASH

The Bond Paid the Bills, Not the Business

Operations used $256.1 million of cash in the September quarter. Counting equipment, free cash flow was negative $279.1 million by our math, and negative $400.0 million across the nine months.

Investors supplied the difference. In August the firm sold $1.15 billion of notes that pay no interest and convert in 2031, money Tsai calls "sufficient to fund our working capital needs for the next couple of years". It repaid its bank loan and raised the dividend 15%. Net of cash, debt now stands at $173 million, where three months earlier borrowing was smaller than cash.

What to watch: operating cash flow in the fourth quarter, when Tsai expects stock to build again first. A repeat of this outflow would use about half the bond proceeds in six months, by our math.

๐Ÿ”ด persistent DISCLOSURE

Risks This Quarter Didn't Answer

Silicon Motion reports as a single segment and describes its product lines only as growth ranges, so several standing risks got no number again. Each item below names the reading that would move it.

  • MonTitan revenue: never disclosed; a dollar figure next quarter would test the seven-customer ramp Kou promises for this year.
  • Ferri against boot drives: still one combined share for both lines; a split would show which engine grows.
  • Customer concentration: no customer named here, and the last half-year filing put the largest at 38% of sales.
  • Research capacity: Kou said "we just really don't have enough R&D resources"; January's spending plan answers it.
  • MaxLinear arbitration: argued in March, with Tsai "still anxiously awaiting the results".

๐Ÿ‘“ Other Themes

persistent DISCLOSURE

Investment Marks Swung Profit the Other Way

Silicon Motion holds shares in two of its customers, BIWIN and Techwinsemi. Those stakes lost $78.0 million of value this quarter, after a gain of similar size in June. The swing explains why reported profit per share fell to $1.91 while the figure excluding one-offs reached $4.35. Long-term investments on the balance sheet dropped to $49.5 million.

new PRODUCT

A New Boot Drive Market for 2027

Kou described a second kind of boot drive, one with its own memory chip beside the processor, which servers use to start up and keep logs. Flash makers have left that niche to chase larger drives, so customers came to Silicon Motion. He puts the market at "a minimum $10 billion by 2030" and expects his own product to ramp in the "second half 2027".

๐Ÿ’ฒ Other KPIs

Accounts receivable (26Q3) $443.5 million
โ‡— accelerating

Customers owe more: receivables rose 120% from a year ago, a little behind the sales increase. Collection took 72 days of sales, against 76 a year earlier, so the drive business is not stretching payment terms. Receivables are money Silicon Motion has billed and not yet collected.

Accounts payable (26Q3) $63.5 million
โ‡˜ decelerating

What the firm owes suppliers fell 39% from June, even as it bought more memory. Suppliers now take their cash sooner than customers hand theirs over, which widens the gap the new bond is filling.

Effective tax rate (26Q3) 25.3%
โ‡— accelerating

Tax took a quarter of pre-tax profit, against 13% a year ago, as earnings shifted from investment gains to operations. Tsai expects about 20% in the fourth quarter. The rate matters more now, because operating profit, not marks on shareholdings, carries the result.

๐Ÿ”ฎ Guidance

Q4 2026 Revenue $588 to $616 million
๐Ÿ … raised from At least about $418 million, the floor left by the full-year target (by our math)
โ‡˜ decelerating

New. The middle of the range points to 7% growth from this quarter, and 116% from a year ago by our math. That sits far above the floor the annual target left, which needed only about $418 million. Growth against last year eases for the first time in five quarters, on a base more than twice as large.

Q4 2026 Gross margin, excluding one-offs 54.0% to 55.0%
โ‡— accelerating

New. Tsai expects 54.5% at the middle, about 0.8 points above this quarter, on "continuing improvement in product mix". Silicon Motion has finished above its own margin range in each of the last three quarters, so mix keeps surprising its own planners. Car products and enterprise drives carry the improvement.

Q4 2026 Operating margin, excluding one-offs 36.0% to 37.0%
๐Ÿ … raised from Above 30% exiting 2026
โ‡— accelerating

Raised. Three months ago Tsai said margins could exceed 30% leaving this year, and the new range starts well above that. By our math the plan implies operating costs excluding one-offs near $108 million, against $120.7 million this quarter. Spending eases because the 4-nanometer tape-out is behind the firm, so the step is leverage and timing rather than cuts.

FY2026 Revenue growth More than 100% year-over-year
๐Ÿ † unchanged from More than 100% year-over-year
โ‡˜ decelerating

Unchanged, and now easily covered. Nine months of sales plus the fourth-quarter range put the year near $1.96 billion, or 121% growth by our math. Kou calls the firm "on track to deliver record annual revenue in 2026, representing more than 100% growth over the prior year". The floor has not moved since July, while the quarterly ranges behind it rose twice.

FY2027 Revenue growth At least 40%
โ‡˜ decelerating

New. Tsai said the order book supports the growth rate above, and he will set out detail in January. Kou says 2027 wafers, test slots and assembly capacity are already booked, so the limit he names is engineering staff rather than supply.

FY2027 Gross margin Low to mid 50% range
๐Ÿ … raised from Long-term model of 48% to 50%
โ‡’ stable

Raised. The published long-term model has sat at 48% to 50% for years, and Tsai now expects the low to mid 50s through next year, because car and enterprise drives price better. He negotiates memory cost on six-month contracts, which he says protects the margin when flash prices move.

โ“ Key Questions

How much revenue does MonTitan bring today?

Seven customers are ramping the enterprise controller and Kou calls it a margin helper, yet no revenue figure exists. Without one, the enterprise case rests on customer counts rather than sales.

What did the memory in the warehouse cost?

Inventory equals ten months of costs and was bought during a shortage. An average cost, with the terms for passing price changes to customers, would show how much margin sits at risk.

When does operating cash flow turn positive?

Tsai targets two quarters between buying memory and collecting cash, but expects inventory to grow again first. A date would tell holders whether the dividend still leans on the bond money.

Is there any decision in the MaxLinear case?

Closing arguments in Singapore finished in March, and the claim seeks a $160 million termination fee. Eight months later, neither the timing nor the size of any award is public.

Will the first quarter fall from the fourth?

Tsai says consumer seasonality becomes more muted as enterprise grows, yet he declined to say whether sales dip in the March quarter. The answer sets the base for the 40% year.