Nehal Chokshi

Charles pointed to DCBBS total-solution differentiation and faster time-to-market across silicon platforms as the value-add.

Charles attributed margin improvement to both DCBBS (with 20%+ margins) and enterprise customer growth. Both factors expected to continue driving improvement.

Charles confirmed the 20% DCBBS target is profit contribution, not revenue. Indicates high-margin DCBBS business targeting meaningful share of total profitability.

Charles deflected the specific customer question by pointing to overall growth trajectory. Noted more diversified large customer base and mentioned $50-60B revenue aspiration.

Charles confirmed SMCI's engineering advantage enables 3-6 month time-to-market lead on total solutions including DLC and dry cooling towers ahead of NVIDIA reference architecture.

David framed margin-dilutive mega-deals as validation of DCBBS capabilities that generate follow-on business. Charles noted xAI remains an important long-term partner.

David cited MI-355X and GB300 ramp as Q1 revenue drivers. Lack of operating leverage attributed to production learning curve during platform technology changeover.

Charles confirmed DCBBS is bundled as comprehensive total solution including computing, cooling, power, battery backup. Expects DCBBS to reach over 20-30% of deal value over time.

Charles expects inventory reserve impact to decline from 200bp in Q3 to approximately 100bp in Q4 and near zero by September.

David declined to share backlog figures but noted backlog builds with new chip cycles and expects growth industry-wide.

Charles claimed validation from both historical growth rates and customer pipeline, backlog, and commitments, but shared no specifics.

Charles said he is considering it with an open mind and acknowledged he will eventually retire, but gave no commitment or timeline.

David confirmed 10% customers exist but declined to disclose the percentage. Charles added they are gaining new customers in Europe and Asia.

Marc Boroditsky said the 3.5x AI cloud pipeline excludes hyperscaler deals and reflects qualified opportunities across AI natives, ISVs and enterprises.

Arkady Volozh said capital is split between scale (data centers, GPUs) and product, which is built organically and supplemented by acquisitions like Tavily.

Management attributed the drop to the capacity bottleneck and said Q4 incremental ARR will be significantly higher as new capacity comes online.

Dado Alonso said Nebius is evaluating asset-backed financing, corporate debt and equity, expecting attractive terms backed by its largest customers.

Andrey Korolenko said cloud training speed scales linearly and matches bare-metal performance, a key selling point for customers.

Dado Alonso gave concrete ARR progression and said a significant portion of the raised guidance is already under contract.

Singh pushed back on any implied decline, saying combined-company guidance and SkyWater eliminations will be provided next quarter.

de Masi confirmed Lightsynq memory sits in the EPB networking work, while SkyWater builds ever-larger single-chip qubit counts.

Singh deflected to EBITDA as the better metric given heavy R&D, citing product-mix effects and a low bill-of-materials advantage.

de Masi confirmed the walking cat architecture and low physical-to-logical ratio were contemplated in the June roadmap.

Wissam declined to break out AI bookings or revenue, only quantifying the number of on-prem AI wins.

George clarified 5x growth is revenue-based not customer count. Win rates very strong; focus is expanding deal pipeline via refreshed portfolio, NVIDIA certifications, full-stack reference architectures, and expanded hyperscaler/application integrations.

Wissam confirmed Q4 FX impact de minimis (<1%). FY26 FX embedded in guidance at prevailing rates; declined to comment on analyst's 100bps tailwind estimate.

George confirmed scaled-out architecture solutions expected in customer deployments by end of next calendar year.

Management declined a per-quarter megawatt figure but cited a robust pipeline and 2 gigawatts added in 12 months; partial.

Management said the 36-qubit fidelity improvement to 99.6% came mainly from design optimization, with the 9-qubit at 99.8%.