They guided gross margin down 70bps while guiding operating margin up 400bps and revenue up 40% sequential.
Thesis: Scorpio X crossing $1,000 content per XPU on the 320-lane device and hitting volume production one quarter early is a pricing power event, not a cyclical print. The Q3 guide is underwritten by one lead hyperscaler, but multiple platforms at that same customer plus 10+ engaged accounts give a second derivative of upside. Gross margin drift to 70% is the cost of buying share in the $20B switching TAM while opex leverage drives EPS to $1.16-1.21.
Verdict: LONG — Conviction: MEDIUM
Catalyst: Q3 print of the $540-560M guide with Scorpio as the largest product line, then second Scorpio X customer revenue by end of 2026 as stated by Sanjay: 'additional customers ramping to production towards the end of this year, early next year.'
Key Risk: Single-socket dependency. The guide's largest component is one lead customer's 320-lane qualification timeline, which sits with the customer: 'based on the progress that they can make.' A slip pauses 40% sequential growth.
The Tell: Sanjay volunteered, unprompted at the end of a concentration question: 'even with the lead hyperscaler we have, there are multiple platforms where we are designed in. It's not like there is just one, right? I think that's a nuance that needs to be considered.' He raised the single-customer risk before anyone asked, which means he knows the Street is underwriting exactly that.
Friction Level: MODERATE_FRICTION — Both sides accept the Q2 beat and the Q3 guide as real. The disagreement: whether the guide is a diversified content-per-XPU ramp or a single socket that a customer qualification decision can pause.
2026Q2 · Aug 4, 2026
3 of 5 readings agreed3 LONG · 2 HOLDRead the full call
Q3 Guide Is 40% Sequential, Anchored to One Hyperscaler's Platform Ramp
Scorpio X crossing $1,000 content per XPU on the 320-lane device and hitting volume production one quarter early is a pricing power event, not a cyclical print. The Q3 guide is underwritten by one lead hyperscaler, but multiple platforms at that same customer plus 10+ engaged accounts give a second derivative of upside. Gross margin drift to 70% is the cost of buying share in the $20B switching TAM while opex leverage drives EPS to $1.16-1.21.
Sanjay volunteered, unprompted at the end of a concentration question: 'even with the lead hyperscaler we have, there are multiple platforms where we are designed in. It's not like there is just one, right? I think that's a nuance that needs to be considered.' He raised the single-customer risk before anyone asked, which means he knows the Street is underwriting exactly that.
Q2 revenue $392.4M beat with GM 73.7% vs 73% guided, then Q3 guide of $540-560M is 40% sequential with EPS up to $1.16-1.21 from $0.80. Scorpio X entered volume production a quarter ahead of plan.
Content per XPU went from $50-100 at IPO to multiple thousands today, with Scorpio X alone above $1,000 per XPU on the 320-lane device. ASPs rising on PCIe 6 retimer attach without volume loss.
COSMOS moved from telemetry to orchestration: 'the XPUs are talking directly to our COSMOS platform to run the workloads.' Sanjay: customers adopting this generation 'will continue to use this for the next generation.'
GM guides down to 72% yet operating margin guides up 400bps to 43% on revenue leverage. Opex up only $20M against ~$150M sequential revenue adds.
Q3 guide rests on one production ramp at the lead hyperscaler; other Scorpio X accounts are pre-production with revenue 'by end of this year, early next year, based on the progress that they can make.'
72% GM guide attributed to 'wider range of gross margins' and silicon versus module mix, explicitly excluding Scorpio while Scorpio becomes the largest product line. Steering to 70% 'from here.'
TAM claims stacked at 2027+ dates: $20B switching, Taurus doubled to $4B by 2030, tens of billions optical, multi-billion custom silicon. Almost no incremental revenue beyond Aries and Scorpio X this year.
+72
CONFIDENT"No hedging on the guide. Sanjay gave a specific volume date ('ramp to high volume production in Q3'), named configurations, and disclosed over 10 engaged Scorpio X customers unprompted. CFO tied the 72% GM guide to silicon/module mix with a stated 70% long-term target."
Specific revenue, margin, opex, share count, and EPS figures disclosed with product-line detail. No working capital, DSI, receivables aging, or book-to-bill data. Concentration quantified only qualitatively.
Q3 print of the $540-560M guide with Scorpio as the largest product line, then second Scorpio X customer revenue by end of 2026 as stated by Sanjay: 'additional customers ramping to production towards the end of this year, early next year.'
Single-socket dependency. The guide's largest component is one lead customer's 320-lane qualification timeline, which sits with the customer: 'based on the progress that they can make.' A slip pauses 40% sequential growth.
The verdict is public. The case behind it is Pro: the forensic reading of guidance and language, the bull and bear cases, the red flags, and the play. It opens to everyone on Nov 2, 2026.