NVIDIA Engagement Bullets — Draft v0
Per Joe Malchow’s structural guidance (Joe Malchow advice, 2026-06-17): four to five bullets, descending specificity, problem-first, KPI-anchored. Premise → problem → rationale for an arm’s-length third-party repository → concrete KPI.
-
Premise. AI silicon is the backbone of the next decade of the information economy, and the buildout has outrun the service envelope around it. The layer that handles replacement, repair, escalation, and customer-state still runs on tooling built for an era when a single customer relationship was an order of magnitude smaller and the silicon was an order of magnitude more forgiving. NVIDIA is the first OEM to feel that mismatch at scale; it will not be the last.
-
The problem at NVIDIA, more precisely. NVIDIA does not have a coherent view of its commitments to customers, what those customers have in the field, or how either side is performing against the other. Greg, in the conversation: “We don’t even know the score of the game.” Lonny: “We come to the conversation, and we’re not even armed.” The data exists. It lives across SFDC, SAP, the NVIDIA data lake, customer reply-all chains, and 3PL spreadsheets, with no single source of truth and no shared view between NVIDIA and the customer.
-
Why the problem is not solvable from inside NVIDIA’s four walls. A correct solution requires data NVIDIA cannot collect directly: customer-side telemetry on how units are performing in production, contract-manufacturer repair and dispositioning records, 3PL shipment events, distributor inventory state. Lonny was explicit: “You’re not gonna get telemetry off their devices.” Customers strip the chips before they come back because they treat NVIDIA as a partial competitor. An arm’s-length, third-party data repository, operated independently of NVIDIA, is the structural answer. It can collect from the rest of the supply chain in ways no single OEM ever could. NVIDIA’s data remains perpetually NVIDIA’s; the operating company is ours to build and run.
-
Concrete targets, twelve months.
- 95%+ SLA attainment on standard RMAs, measured for the first time.
- Hyperscaler RMA cycle time reduced 30%+ against current baseline.
- Customer-initiated escalations per open RMA cut by half.
- Stranded-shipment incidents (the Phoenix five-truck event) driven to zero.
Anchor docs: NVIDIA latent problem codification; source interview Greg + Lonny @ NVIDIA, 2026-06-17.