NVIDIA

Executive Summary

NVIDIA is the control plane for accelerated computing, combining processors, systems, networking, software and a developer ecosystem into unusually strong platform economics.

• Q2 FY2027 revenue doubled to $96.2B and Data Center reached $89.0B.
• GAAP gross margin was 75.0% and operating margin was 66.2%.
• At $230.36, valuation already capitalizes a long runway of AI infrastructure expansion.

Company History

NVIDIA evolved from PC graphics into a full-stack accelerated-computing company through CUDA, data-center systems and high-speed networking.

• CUDA converted GPU programmability into a durable developer and software moat.
• Mellanox added InfiniBand and Ethernet networking to the compute platform.
• Hopper, Blackwell and Vera Rubin moved NVIDIA from chips toward rack-scale AI infrastructure.

Business Model

NVIDIA monetizes architecture leadership through chips, boards, complete systems, networking, software and services, while relying on a global fabless supply chain.

• Data Center represents more than 90% of current quarterly revenue.
• Hyperscalers and AI clouds purchase integrated compute and networking systems.
• Edge Computing spans gaming, workstation, automotive and local AI markets.

Financial Quality

Growth, margins and cash generation are exceptional, but equity-security gains and working-capital expansion make GAAP net income less informative than operating profit and cash flow alone.

• FY2026 revenue grew 65% to $215.9B and operating income reached $130.4B.
• Q2 FY2027 operating cash flow was $24.1B despite a $22.3B increase in receivables.
• H1 net income included $23.7B of gains from equity securities.

Platform Economics

The moat is system-level: accelerated compute, NVLink, networking, CUDA libraries and deployment tooling raise performance while increasing customer integration costs.

• Q2 Data Center revenue grew 117% to $89.0B.
• Hyperscale revenue was $48.7B; AI Clouds, Industrial & Enterprise was $40.3B.
• Blackwell remained the vast majority of revenue while Rubin entered production shipments in Q3 FY2027.

Supply and Capacity

Long-lead procurement secures growth but transfers demand-forecast risk onto NVIDIA as commitments become larger than annual revenue.

• Supply and capacity commitments rose to $279B from $119B one quarter earlier.
• Inventory reached $31.6B, up from $21.4B at FY2026 year-end.
• FY2026 included a $4.5B H20 excess-inventory and purchase-obligation charge.

Balance Sheet and Capital Allocation

NVIDIA has substantial liquidity and cash generation, while strategic investments, debt issuance and infrastructure support are making capital allocation more complex.

• Cash and marketable debt and equity securities totaled $99.4B at Q2.
• Debt increased to $33.4B after a $25B senior-note issuance.
• NVIDIA repurchased $39.0B of shares in H1 FY2027 and retained $99B of authorization.

Growth Drivers

Demand is expanding from model training into inference, agents, sovereign AI, enterprise workloads, networking and physical AI.

• Q3 guidance calls for $108B of revenue without China Data Center compute revenue.
• Rubin provides the next architecture transition after Blackwell Ultra.
• Networking, enterprise adoption and software can broaden revenue beyond hyperscale training clusters.

Competition

NVIDIA competes against custom accelerators, AMD, merchant networking, emerging domestic Chinese platforms and customers seeking lower total cost and supplier leverage.

• Hyperscalers can shift workloads to internally designed ASICs.
• Open software layers may reduce CUDA switching friction over time.
• Architecture transitions require simultaneous execution across compute, memory, networking, cooling and systems.

Management and Governance

Founder-led technical strategy has produced extraordinary results, while expanding investments and guarantees require increasingly disciplined disclosure and capital governance.

• Jensen Huang remains central to product vision and ecosystem relationships.
• Strategic stakes may strengthen demand but can blur customer financing and investment returns.
• Compensation, succession and related ecosystem commitments deserve continued monitoring.

Risks

The largest risks are an AI capital-spending correction, customer concentration, export controls, supply mismatches, competition, and contingent infrastructure exposure.

• Two direct customers represented 22% and 14% of FY2026 revenue.
• NVIDIA says it is effectively foreclosed from China's data-center compute market under current rules.
• Q2 disclosures include $36B of AI-cloud commitments, $20B of third-party leases and material guarantees.

Valuation

A strong business can still be a weak investment if growth duration and terminal margins are overestimated; the current price requires sustained scale and cash conversion.

• The bear case assumes demand normalizes and FCF margin falls toward 24%.
• The base case assumes a long AI build cycle and high-30s mature FCF margins.
• The bull case requires NVIDIA to remain the dominant full-stack platform through multiple architecture generations.

Investment Conclusion

NVIDIA has the strongest demonstrated economics in AI infrastructure, but investors are underwriting both technological leadership and the solvency and utilization of an increasingly financed ecosystem.

• Track Data Center growth and gross margin together, not accelerator demand alone.
• Monitor inventory, receivables, supply commitments, cloud agreements and guarantees as a single risk system.
• A durable thesis requires Rubin execution and widening enterprise inference demand after hyperscale growth matures.

$61

12.0% discount · 3.0% terminal growth

$192

10.5% discount · 3.5% terminal growth

$419

9.5% discount · 4.0% terminal growth