Semtech Sets FY30 Targets of $3.2 Billion Revenue and $2.1 Billion From Data Center, With 70% Gross Margin and $12.40 to $13.00 EPS
Data Center Analyst Day, October 15, 2026: Q3 FY27 outlook raised as management ties a multiyear model to record backlog
Semtech used its first Data Center Analyst Day, held alongside the OCP Global Summit, to publish its first multiyear financial framework. By FY30 the company targets revenue of $3.2 billion, a 30% CAGR from FY26 that excludes $166 million of cellular module revenue slated for divestiture in Q4. Data center revenue is expected to compound at 75% to $2.1 billion, roughly two-thirds of the total. Adjusted gross margin is guided to 70%, adjusted operating margin to 48% to 50%, and free cash flow to 35% of revenue, or more than $1.1 billion. Adjusted EPS is targeted at $12.40 to $13.00, a 64% to 66% CAGR, on 104 million diluted shares and a 20% non-GAAP tax rate. CFO Mark Lin also raised the Q3 FY27 outlook.
The Near-Term Numbers Are Backed by Backlog
The raised Q3 outlook midpoints are revenue of $417.5 million (up 56% year over year), consolidated adjusted gross margin of 59%, a record adjusted operating margin of 33%, and a record adjusted EPS of $1.14, up 138%. Data center is expected to reach $145 million in the quarter, about 40% of revenue, versus $61 million for all of FY24. LoRa is expected at $68 million, or 19% of revenue.
Investors should read the margin optics carefully. The 64% adjusted gross margin excluding the held-for-sale cellular modules business implies a 950 basis point sequential step. Much of that is the mechanical removal of a lower-margin business, not organic improvement. Management says the divestiture should close in Q4, and the underlying mix shift toward data center and LoRa is real. The operating margin gain is more organic: adjusted opex is guided to 26% of revenue, down 910 basis points from FY25, even as R&D dollars rose 27% over the same period.
The more consequential claim is visibility. CEO Hong Hou said booking momentum has produced record backlog for four consecutive quarters. He said no customer has asked to dial demand down, some are trying to pull in deliveries scheduled for next year, and the company sees no sign of channel inventory build. Lin said data center revenue should more than double in FY27 and again in FY28, "grounded in backlog," then layered with design wins that have yet to ramp. Those two years imply data center revenue of roughly $1 billion by FY28, which leaves much of the FY30 target reliant on a slower second half of the plan.
The Outer-Year Deceleration Is Deliberate
The model front-loads growth. Hong Hou called the plan "heavily indexed on the near term" because that is where visibility exists. In FY29 and FY30 the company assumes growth at the market CAGR. One analyst asked directly about the implied slowdown. Lin said outer-year growth is "grounded in market CAGR assumptions" and that if the market proves larger, Semtech would capture it. Management declined to split the $2.1 billion between pluggable modules and newer NPO or CPO formats, and said it had not baked in anything close to the potential content gains from lasers.
The company's serviceable market framing supports the ambition. Semtech sizes its SAM at $3 billion growing to $15 billion over four years, using 650 Group data. Management said its own growth should run about 50% above the market. The three drivers are market growth, content per module, and share gains on future line rates.
Content Per Module Is the Core Thesis
Hong Hou said Semtech's content at 800G is high single-digit dollars per module. The company sees a path to more than 10x that as it adds CW lasers, photodiodes and eventually modulators at 3.2T. A slide referenced by Craig Ellis showed content moving above $100. Imran Sherazi illustrated the scale with a hypothetical data center of 500,000 optical modules.
The company's share position is the evidence. Sherazi said Semtech's 100G-per-lane TIA took share from 18% to roughly 50%, and that the 200G generation is only beginning. Management also said its 336G linear redriver is "the only solution out of 6 competitive re-drivers" that currently performs in an AI data center application, a customer-validated proof point. Sherazi said the company added 20 dB of link budget to an ASIC at 224G with a major AI supplier, and Hong Hou said linear approaches cut power by up to 90% versus retimers.
Hong Hou was careful about scope: "We don't do cooling or power delivery. We don't do compute. We don't do memory either, but we're just laser focused on providing the best network connectivity solutions." He also pushed back on any terminal-value call for copper: "Use copper when you can, use fiber when you need to." Sherazi framed the product logic as "linear when you can and retime when you must."
Joe Moore of Morgan Stanley asked whether active copper adoption is smaller than expected. Management gave no adoption data. Sherazi said the format is irrelevant because "the channel is the channel." That answer is consistent with the strategy but leaves the copper revenue trajectory unquantified.
Photonics Expands the Content Story but Is Capacity Constrained
The March acquisition of HieFo, now Semtech Photonics Products, and the June acquisition of D-Tech add indium phosphide lasers and high-speed photodiode design. Management said demand exceeds capacity today. The plan is to raise capacity 3x to 4x by the end of the year, helped by an existing 7,000-square-foot clean room obtained with D-Tech. Samples and a 300G photodiode demo are planned for OFC.
The company cited 44% power conversion efficiency, coupling loss below 0.8 dB versus 1.2 to 1.5 dB typical, and a saving of 40 to 50 milliwatts in a 100G module. Hong Hou said early customer results show the lasers outperforming incumbents. Management is not chasing commodity volume: "We're not going to be competing with the commodity type like $3 per 30-milliwatt lasers." It is targeting very high-power CW lasers of 150 to 250 milliwatts, where one laser can drive 8 channels, plus tunable laser gain chips for scale-across and, later, OCI scale-up sources. Hong Hou acknowledged the portfolio is "still limited" and relies on partners for modulators. The US-made supply chain is a selling point.
A multiyear supply agreement with Tower Semiconductor, announced Tuesday, secures silicon germanium capacity beyond FY30, which addresses the supply-assurance risk behind the data center ramp.
Industry Panel: CPO Is Later Than the Hype, NPO Comes First
A panel moderated by Morgan Stanley's Mark Edelstone added external context. Alan Weckel of 650 Group said scale-out is the first networking domain to cross $100 billion, and scale-up will eventually be larger. He estimated networking at 12% to 15% of spend per gigawatt, rising toward 20%, against total spend of $35 billion to $50 billion per gigawatt. He placed the CPO inflection in the second half of 2028 or 2029, gated by manufacturing test and qualification and by switch and XPU roadmaps.
Hong Hou aligned with that view. He said NPO volume starts around the end of next year, delivering roughly 85% of CPO's power benefit while leaving the XPU socket open. CPO follows NPO. The beneficiary case for Semtech is that NPO builds on 200G-per-lane LPO designs already validated in pluggables. Arista's Sunil Priyadoshi presented the XPO pluggable form factor, which packs eight 1.6T modules into 12.8T, supports linear operation, and has drawn 175 MSA members. Semtech says it is form-factor agnostic.
On the switch ASIC vendors' potential encroachment, Chief Commercial Officer Jason Green said customers see Semtech as a trusted partner, not a future competitor, because it stays in analog connectivity.
Balance Sheet Repair Is Largely Complete
Lin said net leverage fell from 10.5x in FY24 to 1.1x, with net debt of $299 million ($503 million debt, $204 million cash) and total liquidity of $561 million. Net cash interest cost $82 million in FY24, and free cash flow was negative $123 million. Trailing free cash flow is now $193 million. Maturities are light: $100.5 million in FY28 and $402.5 million in FY31. "Over leverage is in our history, but it is history," Lin said. Management targets a maintenance net leverage ratio below 0.0x, meaning net cash.
Hong Hou was candid about the cause, calling the 2023 Sierra Wireless acquisition a deal done "at the wrong time, which stretched our balance sheet and diluted our focus." Capital priorities are customer-aligned R&D first, then capability-driven M&A at various scales, then balance sheet strength. Asked about buybacks and dividends against more than $1.1 billion of FY30 free cash flow, Lin did not commit to any return program.
LoRa Is a Second Growth Leg, With Sidewalk Treated as Upside
LoRa revenue, stuck near $150 million annualized, is on track for $250 million to $260 million this year. Management guides a 30% CAGR through FY30 across smart building, security, drones, edge AI and medical. Gen 4 delivers 2.4 to 2.6 Mbps, and Gen 5 targets 6.4 Mbps with 2 to 3 dB better sensitivity and lower standby current. Hong Hou said the plan includes only a "very moderate" contribution from Amazon Sidewalk, which could prove a meaningful swing factor if consumer scale develops.
The 2030 targets are ambitious and sit on a very sharp inflection, with data center up from 7% of revenue in FY24 to 40% in the coming quarter. Execution risk lies in ramping photonics capacity, defending share as 200G and 400G generations arrive, and the market's own choppiness. Hong Hou acknowledged the week had been exhausting because "the market has not been very kind." Semtech has met or exceeded its revenue and EPS midpoints for 10 consecutive quarters, and that record is the main basis for taking the multiyear model seriously.