SITM · SiTime Corporation

Current Thesis

The MEMS precision-timing picks-and-shovels thesis is no longer a question mark the Q1 2026 print (reported ~2026-05-07) confirmed it emphatically. Revenue of $113.6M (+88.3% YoY) beat the ~$103M consensus, and the Communications/Enterprise/Datacenter (CED) segment grew 158% YoY to $75.7M, roughly two-thirds of total revenue. That blows past the 25%-datacenter-mix bar that the prior framing set as the bull confirmation. Management raised FY2026 revenue growth guidance to at least 80% and guided Q2 to $140–150M. The narrative accelerated exactly as the bulls modelled: AI-inference bandwidth lifting both timing content per box and ASP, with 1.6 Tb optical modules and co-packaged optics (CPO) designs named as drivers.

The catch is the entry, not the story. The stock has re-rated to roughly $700 (anchored by the CEO's 2026-06-03 open-market sale at $701.13) and now trades above the ~$600–625 analyst price-target cluster. The binary that drove the move is spent, the next company catalyst is six-plus weeks out, and the C-suite is selling into strength. A high-quality narrative bought at a stretched price above every target, into a catalyst vacuum, is a probe at best the right exposure is built on a pullback-and-rebase or on Q2 confirmation, not on a chase at all-time highs.

Bull Case

• Q1 2026 (reported ~2026-05-07): revenue $113.6M, +88.3% YoY vs $60.3M a year ago; beat the ~$103.45M consensus. EPS $1.44 vs $1.17 estimate a 23% surprise.
• CED/datacenter segment $75.7M, +158% YoY, now ~67% of total revenue. The datacenter-timing story is now disclosed company revenue, not sell-side TAM math.
• FY2026 revenue growth guidance raised to ≥80%; Q2 2026 guided to $140–150M, roughly +30% sequential at the midpoint.
• Gross margin 64.5% vs 57.4% YoY datacenter-grade SKUs (Elite 2 Super-TCXO) carry richer mix; 1.6 Tb optical and CPO ramps cited as the demand engine.
• Balance sheet: $789M cash post-quarter, augmented by a May 2026 convertible senior notes offering funding capacity for the capacity build-out.
• Sell-side re-rated up: the street-high moved from Stifel's $500 (2026-04-16) pre-print to a ~$600–625 cluster after the print (TipRanks average ~$600, WallStreetZen $624), confirming the narrative-acceleration leg.

Bear Case

• Price (~$700, per the 2026-06-03 CEO sale at $701.13) trades above the ~$600–625 street PT cluster the stock has outrun even its raised targets, the mean-reversion zone the playbook flags.
• Clustered insider distribution: CEO Rajesh Vashist sold 40,000 shares @ $701.13 (2026-06-03); EVP Engineering Fariborz Assaderaghi sold ~$1.41M across 2026-06-03/04. C-suite selling into a ~40% post-print run.
• The catalyst that drove the re-rate is in the rearview; no company-specific binary until the Q2 print (~late July/early Aug). A six-week-plus vacuum into an extended tape.
• Concentration risk magnified: with CED at ~67% of revenue, the AI-datacenter capex cycle is now the entire story the diversification cushion is gone, and a single hyperscaler or optical-module push-out dents the +158% growth line fast.
• The May 2026 convertible notes carry dilution overhang and read as a "raise at the high" signal that frequently coincides with local tops.
• A good Q2 number is largely priced after an 80%+ FY guide; the asymmetry into the next print skews toward an air-pocket on any deceleration.

Setup & Price Structure

• Price zone roughly $680–720, anchored by the 2026-06-03 CEO open-market sale at $701.13. The post-Q1 breakout shelf sits near $550–560; price is extended well above its own base and above the rising medium-term trend.
• Above the ~$600–625 analyst target cluster the "stretched, above target, no fresh catalyst" configuration that historically mean-reverts before the next leg.
• A constructive re-entry needs either a pullback that holds a higher low above the ~$560 breakout shelf and re-bases, or a flag built into the Q2 print. Strength alone is not the setup here because the catalyst is spent and distribution is visible.
• Required confirmation before sizing up: CED mix holding ≥60% sequentially, no further escalation of insider sales, and price reclaiming highs from a base rather than a vertical extension.

Catalyst Calendar (next 30 days)

• No confirmed company-specific binary inside the window (through ~2026-07-07). The Evercore Global TMT fireside (CFO Beth Howe) already passed on 2026-06-03.
• ~2026-07-30 (est.): Q2 2026 earnings the governing binary. Guide is set at $140–150M, so the bar is explicit. (Outside the 30-day window but the next decision point.)
• Optical/interconnect peer prints and updates through the window (CRDO, AVGO, MRVL, COHR) read across as datacenter-timing demand tells.
• Watch SEC Form 4 filings for follow-on insider sales beyond the June cluster, and any convertible-notes settlement mechanics.

What Would Change Our Mind

• Bull re-arm: a clean pullback to the ~$560 breakout shelf that holds as a higher low, then a flag-and-breakout into the Q2 print that re-establishes a top-tier datacenter-timing setup at a sane entry rather than a chase.
• Invalidation: Q2 2026 revenue (reported ~late July) below the $140M guide floor, or a sequential decline in CED/datacenter revenue, or a weekly close back under the ~$560 post-Q1 breakout shelf any of these signals the re-rate is fully priced.
• Distribution-top tell: insider selling escalating beyond the June cluster while price fails to make new highs.

Correlation Notes

• Tightly coupled to the AI-datacenter and optical-interconnect complex CRDO, AVGO, MRVL, COHR, LITE. The shared driver is the 1.6 Tb optical and CPO ramp, where low-jitter precision timing is content that scales with bandwidth.
• High beta to broad semis (SOX/SMH) and to AI-capex sentiment; a hyperscaler capex-guide cut would hit the whole cohort, and SITM's ~67% datacenter mix means it would absorb the hit disproportionately.
• As the only pure-play public MEMS-timing name, there is no direct comp to fade against; the cross-read comes from optical-module suppliers and hyperscaler capex commentary rather than a peer timing vendor.