NVIDIA Corporation (NVDA) stock analysis
Technology · Semiconductors
From the 10-K filed 2026-02-25
Business type: Franchise
How it makes money
NVIDIA earns primarily by supplying Data Center compute and networking platforms to cloud service providers, AI model makers, and enterprise customers.
NVIDIA operates across two primary segments: Compute & Networking and Graphics. Compute & Networking accounts for the majority of revenue growth, driven by Data Center accelerated computing and networking platforms, full-stack systems like Blackwell and Hopper, Grace CPUs, NVLink interconnects, and AI enterprise software.
Customers include major cloud service providers (CSPs), AI foundation model creators, consumer internet platforms, and global enterprises. They purchase rack-scale AI infrastructure to train and deploy complex neural networks, generative AI, and agentic reasoning systems.
What protects it: the moat
Over 7.5 million developers globally use its CUDA software stack, creating high developer switching costs across its unified programmable GPU architectures.
NVIDIA's key moat is its proprietary full-stack platform architecture, anchored by the CUDA development platform introduced in 2006. By maintaining a single unified programmable architecture across GPUs, software libraries, algorithms, and domain-specific SDKs, NVIDIA enables software developers to build applications that run seamlessly across its hardware family.
Because over 6,000 applications and millions of developers are optimized specifically for CUDA and domain stacks like NVIDIA AI Enterprise, DRIVE, Clara, and Omniverse, switching to competing hardware platforms imposes substantial software rewrite and operational costs on customers.
What growth costs
NVIDIA relies on heavy reinvestment, spending over $76.7 billion in research and development since inception alongside extensive non-cancellable supply commitments.
To maintain a rapid one-year architecture release cadence (moving from Hopper to Blackwell, Blackwell Ultra, and Rubin), NVIDIA must continuously invest heavily in R&D across silicon, hardware design, networking systems, algorithms, and software. More than half of its engineers work on software.
Because NVIDIA operates a fabless manufacturing model, sustaining growth also requires significant cash outlays in advance. To secure specialized wafer fabrication capacity, advanced packaging like CoWoS, and critical raw materials, NVIDIA must enter long-term supply agreements, pay premiums, make non-cancellable purchase obligations, and provide deposits.
What could damage it
Geopolitical trade restrictions and export license mandates effectively foreclosed NVIDIA from China's data center market, causing a $4.5 billion inventory charge.
An expanding set of export control regulations by the U.S. government restricts shipments of high-performance compute products (such as A100, H100, H20, and H200) to China and specified regional markets. When licensing rules changed for the H20 integrated circuit in fiscal year 2026, demand collapsed, forcing a $4.5 billion charge for excess inventory and purchase commitments.
Being foreclosed from competing in China's data center compute market permanently harms earnings by severing access to major revenue streams while encouraging Chinese domestic customers and developers to build alternative hardware and software ecosystems.
About NVIDIA Corporation
NVIDIA Corporation operates as a data center scale AI infrastructure company in the United States, Taiwan, China, Hong Kong, Europe, and internationally. It operates through Compute & Networking, and Graphics segments. The Compute & Networking segment provides data center accelerated computing and networking platforms and artificial intelligence solutions and software, and automotive platforms and autonomous and electric vehicle solutions, including software. The Graphics segment offers GeForce GPUs for gaming and PCs; Quadro/NVIDIA RTX GPUs for enterprise workstation graphics. The company's products are used in gaming, professional visualization, data center, and automotive markets. It sells its products to original equipment manufacturers, original device manufacturers, system integrators and distributors, independent software vendors, cloud service providers, add-in board manufacturers, distributors, automotive manufacturers and tier-1 automotive suppliers, and other ecosystem participants. It has a collaboration with Cisco to develop Ultra-Dense Small Form Factor Pluggable (USFP), a high-density optical module pluggable transceiver. It has a collaboration with ChipAgents to develop Build-A-Chip-Bench, a benchmark for AI agents carrying chip design tasks and The Company has a strategic alliance with MediaTek to build the next generations of AI computing platforms. NVIDIA Corporation was incorporated in 1993 and is headquartered in Santa Clara, California.