NVIDIA's $96.2B quarter decoded by an AI finance model: $36B cloud deals are promises, not revenue
anthara_ai · x · 2026-09-17
A blogger used InclusionAI's finance-focused model Ling-3.0-flash-fin to parse NVIDIA's FY2027 Q2 filings under strict rules: cite only the PDFs, flag evidence for every judgment, and say 'cannot infer' when it can't.
Booked results: quarterly revenue of $96.221B (+106% YoY), data center revenue $89.023B (+117%), GAAP net income $59.688B, and $74.421B operating cash flow over the first six months.
A second set of headline numbers belongs elsewhere: $36B in AI cloud commitments, up to $105B in SB Energy guarantee caps, and a platform arrangement aiming to unlock over $500B in third-party capital—future commitments, guarantees, and proposals that shouldn't be counted as quarterly revenue. The model's practice of separating confirmed / in-progress / undeterminable items is highlighted as genuinely useful.
Related event: Analysts Use Financial AI to Break Down NVIDIA's $96.2B Quarter(2 posts)→
More from Infra
- ROCm Beats Vulkan on Strix Halo: Up to 47% Faster Decoding at 41.9 tok/s — QuixiAI · 2026-09-17
- Qwen3.8-Flash on 4x AMD V620 hits 1,300 PP and 70+ tok/s on coding — Thin_Pollution8843 · 2026-09-17
- Retraining fp16 scales only: patching a 3-bit Qwen3.8-27B GGUF closer to its BF16 parent — ZenZombie117 · 2026-09-17
- New dashboard sets first public baseline for advanced RL training costs — teortaxesTex · 2026-09-17
- PyTorch Conference NA lineup spotlights torch.compile and custom kernel breakthroughs — PyTorch · 2026-09-17
- Structured-decision trick speeds up DiffusionGemma inference 3-10x with one forward per request — bodonoghue85 · 2026-09-17