Earnings Highlights: SiTime (SITM) | Q2 FY2026 — Acquiring Renesas' Timing Business Puts Precision Timing on the Road to a Billion-Dollar Club

2 days ago

7 min read

This article is for technology and industry trend analysis only and does not constitute investment advice.

- This quarter is an inflection point in financial quality, not just growth. Revenue was $157.4M, up 127% YoY, but the real signal is that operating margin jumped from 10.3% a year ago to 34.0% (non-GAAP), with gross margin rising to 67.1%. Revenue doubling while margins jump at the same time means this is scaling, not a one-off order padding the numbers.

- CED (Communications, Enterprise & Data Center) is the engine, and it isn't slowing down. The segment did $101.2M in the quarter, up 181% YoY, formally crossing $100M per quarter — its ninth consecutive quarter of triple-digit growth. 1.6T optical modules, data center synchronization, and AI data centers spreading beyond traditional cloud players are all firing at once.

- SiTime closed the acquisition of Renesas' timing business (TPD) on July 1, and Q3 guidance jumps to $285–295M. Management originally targeted $300M within 12 months of closing; now it hints at "higher." The deal moves SiTime from a "MEMS oscillator company" to a full-line timing supplier spanning oscillators, resonators and clock ICs — and pulls forward the timeline to $1B in revenue.

1. Three Signals to Remember First

First, the margin jump matters more than the revenue doubling. A semiconductor company growing revenue 127% YoY isn't unusual; what's unusual is operating margin going from 10.3% to 34.0% in the same quarter, with net margin above 40%. It shows SiTime's product mix is moving to the high-value end (high-end TCXOs, Super-TCXOs) rather than trading price for volume.

Second, the strength this quarter is broad-based, not a single point. Management stressed that every BU and every region grew more than 50% YoY, some over 100%. When only one engine is running, you worry; when all of them run together, the signal becomes "end demand has really arrived."

Third, visibility is extending into 2027. Customers are already ordering 12 to 18 months ahead, a strong order-behavior signal in the passive/timing component industry. Broadcom's results pointed the same way on extended AI order visibility; we broke down the same logic in Optics Replacing Copper, This Time Backed by Earnings: Broadcom Extends Visibility to 2027.

2. Revenue and Financials

Operating expenses (non-GAAP) were $52.1M, including R&D of $25.6M and SG&A of $26.5M — expenses are rising, but revenue is growing much faster, which is operating leverage. On the balance sheet, operating cash flow doubled to $40M (vs. $15.3M a year ago) and free cash flow was $27.1M; inventory rose to $103.9M ahead of Q3 shipments. In addition, the $1.35B zero-coupon convertible notes (due 2031) issued in May contributed $12.2M of interest income this quarter, but that cash was used for the acquisition consideration in July, so interest income drops back to about $4M next quarter — a one-off sweetener to strip out when reading Q3 results.

3. Technology and Business Highlights: Three Drivers of CED

Management broke CED's momentum into three drivers, each tied directly to the physical limits of AI infrastructure.

The first is bandwidth upgrades. As optical modules move to 1.6T, SiTime expects its 1.6T revenue to grow 100% in 2027, with 800G also growing significantly; together they map to a 2027 SAM of about $450M, where SiTime holds significant share. The key point: each generational jump raises clock frequency and jitter requirements, squarely in the range of SiTime's high-end products. For how far 1.6T/3.2T pushes clock specs and why the fs numbers on a datasheet can mislead, see our full breakdown in The fs on the Datasheet Can Lie: Understanding the 1.6T/3.2T Clock Bottleneck Through Six 312.5 MHz-Class Clocks; for how the jump to 1.6T redistributes BOM value, see 800G vs. 1.6T Optical Modules: A BOM Teardown of Who Really Captures the Margin.

The second is the spread of synchronization. Hyperscalers are starting to deploy synchronization across compute and network nodes, lifting demand for SiTime's Elite series Super-TCXOs. Management gave a very specific figure: this adds several hundred dollars of timing content per data center rack. It's not about selling more of the same parts, but about raising the timing value per rack.

The third is AI data centers spreading beyond traditional hyperscalers. New OEMs/ODMs, enterprise builds, neoclouds and sovereign data centers are all adopting SiTime, creating demand that "didn't exist before." How the battleground shifted from the demand side to the supply side after roughly $725B of CSP capex was locked in, we discussed in After $725B of Capex Is Locked In, Optical Communications' Deciding Battle Has Moved to the Supply Side.

Beyond CED, SiTime is laying long-term groundwork for "AI leaving the data center": automotive (autonomous-driving positioning accuracy, a $400M SAM, pitching 10x positioning accuracy), defense assured PNT (local timing backup when GPS is jammed or spoofed, a $400M SAM with retrofit opportunities on existing platforms), and personal AI devices / smart glasses / wearables (the MICBU funnel already exceeds $1.2B). These aren't major revenue contributors yet, but they raise the SAM ceiling considerably.

4. Management Outlook: Q3 Guidance and the TPD Deal

Q3 FY2026 guidance (including TPD):

- Revenue of $285M–$295M. Broken down, SiTime standalone is $200–210M (+30% QoQ at the midpoint), with the newly acquired TPD at about $85M. - Gross margin about 68% (±1pt), operating expenses $80–85M, interest income about $4M, share count about 32.8M (including 3.6M new shares issued for the acquisition). - Non-GAAP EPS of $3.50–$3.65.

Two pieces of wording are worth unpacking. First, standalone +30% QoQ is a "step change"-level jump; management attributed it to backlog strength and customers' confidence in their own demand (especially in CED), and the tone was "confirmation" rather than "expectation" — they have firm orders in hand. Second, TPD is clearly running ahead. The original guidance was $300M of revenue in the 12 months after closing (about 40% growth over TPD's 2025), yet Q3 alone is pegged at $85M, already annualizing well above $300M. The CFO said outright, "we're on a path to exceed $300M."

TPD's profile (Renesas' former timing business, with roots in ICS → IDT → Renesas) also explains why the deal is a good buy: roughly 70% gross margin, 10,000 customers, and nearly 70% of revenue from CED. The FemtoClock and VersaClock clock families have been around for 20–25 years, and even the buffer line, seen as low-differentiation, generates over $100M in revenue. Its growth drivers are the same as SiTime's own CED — AI data centers. In other words, SiTime isn't buying a stable but stagnant legacy business; it's plugging a clock franchise that also rides the AI wave, but whose growth was underestimated, into its own channels and supply chain.

5. Supply Chain and Customer Clues

A few signals hidden in the details:

Supply during the TSA period is a short-term variable. TPD still relies on a transition services agreement (TSA) with Renesas for manufacturing, test and supply chain, and SiTime will depend on Renesas as a foundry for the next several quarters. Management said plainly that "customers are seeing quite a bit of supply tightness" and that it will take a few quarters to improve — meaning TPD's near-term growth ceiling is supply, not demand, an execution risk to watch next quarter.

A large consumer customer's next-generation design is ramping. MICBU's second-half growth comes mainly from the large consumer customer, which contributed $22.8M this quarter, deploying SiTime products in its next-generation platform. Management's phrasing — "there's a lot of information in the market about how fast they're rolling out" — effectively hints that this customer's product cycle will show up directly in Q3.

The Bosch quartz contract expires next March; management says renewal is "zero issue." TPD brings in a small quartz-based oscillator business, but the clock ICs themselves don't use MEMS resonators, so the acquisition doesn't affect the Bosch relationship, and renewal is expected to be wrapped up soon.

The next technology narrative is "timing moving from discrete components into the system core." Management pointed future direction toward chiplets, advanced substrates and modularization — integrating timing into the heart of the system, which opens an additional $2.5B of SAM in CED alone by 2030. ASPs will rise, but more importantly "usage density" increases: switches and accelerator cards across the rack all need more precise frequencies, and signals can't lose synchronization at any stage.

6. Conclusion

SiTime accomplished two things at once this quarter: it proved its model can expand margins while scaling, and used an acquisition to upgrade itself from a single-product-line company to a full-line timing supplier. For those tracking it, the story has shifted from "MEMS replacing quartz" to "precision timing as the invisible toll booth of AI infrastructure."

Three metrics to watch next quarter (Q3 FY2026):

1. Progress on TPD supply — is the $85M set by demand or capped by supply? Whether capacity loosens during the TSA period decides whether TPD is an upside surprise or a drag.

2. Whether standalone CED sustains triple-digit growth — management says Q3 CED will still grow triple digits YoY; this is the foundation of the entire growth narrative.

3. Whether gross margin holds at 67–68% — a higher consumer mix in the second half is a margin headwind; whether the CED and TPD product mix can offset it is the litmus test of whether the margin story is sustainable.

In one sentence: this is no longer a company betting on whether MEMS beats quartz, but a platform company turning timing into must-have content for AI infrastructure — and starting to use acquisitions to accelerate toward $1B in revenue.