NVIDIA CORP (NVDA) stock intrinsic value fair value estimate & owner earnings analysis

NVIDIA designs the graphics and AI accelerator chips (GPUs) and the surrounding software and networking (CUDA, NVLink, Spectrum-X) that have become the dominant computing platform for training and running artificial intelligence models. What was a gaming-and-workstation chip business as recently as fiscal 2023 (year ended January 2023, revenue of about $27 billion, largely flat that year as crypto-mining demand collapsed) has been transformed by the buildout of AI data centers: revenue reached $61 billion in fiscal 2024, $130 billion in fiscal 2025, and $216 billion in fiscal 2026, with cash generation growing even faster than revenue. Because this is a different business than the one that existed before fiscal 2024, we anchored our valuation on the three most recent fiscal years (2024-2026) rather than a mechanical ten-year average, which would be dominated by the old, much smaller gaming-cycle business and would understate what the company can earn today. Using the middle of those three years as a conservative starting point (about $2.74 of owner cash earnings per share, deliberately below the $4.46 actually earned in the just-completed fiscal year), we built a ten-year forecast and then a view of the decade after that. Owner cash earnings are built from operating cash flow, with stock-based compensation and normal replacement-level capital spending subtracted back out, and the swings from AR/inventory/payables growth (which have been consuming a great deal of cash as the business scales) added back so the figure reflects durable earning power rather than one period's cash timing. Growth capital spending -- the portion of capital budgets funding expansion rather than maintaining existing capacity -- was estimated from the ratio of capital spending to revenue over the last five years and excluded from the deduction, since expansion spending is discretionary, not a cost of staying in business. In fiscal 2024 the estimate implied that effectively all of that year's modest capital spending was growth-related. For the growth path, we note that owner earnings per share already grew about 63% in the fiscal year just completed, and the two quarters reported since then (through July 2026) show revenue growth accelerating further, above 95% year over year, as customers ramp the newest Blackwell-generation systems. Despite that, our base case assumes first-year growth of 55%, below the trailing pace, to stay conservative in the face of a business this dependent on a handful of very large customers continuing to expand their spending. Growth then decelerates steadily to roughly mid-single digits by year ten, reflecting a maturing, still-growing but far more competitive market. Our base case assumes NVIDIA's software ecosystem and full-stack systems approach keep it the leading AI compute platform, but that the current spending intensity from a small number of large cloud customers cannot continue at today's pace indefinitely, and that competitors' own chips (several large cloud providers are building their own AI chips) increasingly compete for the more standardized parts of the market. It values the shares at roughly $206. The bear case (roughly $63/share) assumes an early pause: the AI infrastructure buildout enters a digestion period in fiscal 2028 -- echoing the flat fiscal 2023 -- as a couple of very large customers slow spending after the current wave of construction, gross margins compress as internally-developed competing chips erode NVIDIA's pricing power on more commoditized workloads, and the business settles into a lower, low-single-digit-growth, more ordinary semiconductor company for the long run. Continued export restrictions on advanced chip sales to China also permanently remove a growth avenue this case assumes is not recovered. The bull case (roughly $355/share) assumes the current backlog and expansion continues at a strong pace through fiscal 2027 substantially in line with the two quarters already reported, that new categories of AI use (enterprise deployments, sovereign AI projects, and increasingly complex AI reasoning that demands more computing per task) extend the growth runway for several more years beyond what the base case assumes, and that the software and full-system moat continues to command premium pricing for longer. An important caveat: because this is a fast-compounding business, a large share of all three case values -- roughly two-thirds in the base and bull cases -- comes from what we assume the company earns in the years beyond our explicit ten-year forecast, not from the next decade itself. That tail rests on the judgment that a modest 1% sustained annual growth rate is achievable once today's product cycle matures, funded by the substantial research spending NVIDIA already expenses every year rather than by additional capital most companies would need to raise; readers should treat that as the single largest and most uncertain assumption in this valuation, not a minor technical detail. Given the wide range between our bear and bull outcomes, the margin-of-safety price sits at a substantial discount to our base-case estimate, around $104/share, reflecting the unusually wide range of plausible outcomes for a business this early in a technology cycle and this concentrated in a small number of large customers.