A top-ten TGI rank built on a decade of compounding price growth, and — in May 2026 — a 2,400% dividend hike that moves the dividend from rounding error into something the framework can actually score against.
Editorial published July 7, 2026
NVDA sits inside the top ten on the Total Growth ranking across a watchlist of roughly 900 companies, supported by strong Price Growth and Dividend Growth components. The Healthy Income score is far weaker, a direct consequence of the dividend yield staying near zero for most of the past decade. The pattern is unmistakable: the system is reading NVDA as a pure compounder, not an income stock, with the price-growth side doing the work. One timing note matters — the ranking snapshot behind this page predates the board's May 20, 2026 decision to raise the quarterly dividend from $0.01 to $0.25 (a 2,400% increase), so the income-weighted strategies will re-rate as that new dividend feeds through the four-year average yield in subsequent cycles.
The four TGI ranking scores, current price, and growth figures are on the Stock Lookup — use the tabs above to move across.
| Sector | Information Technology |
|---|---|
| Industry | Semiconductors |
| Exchange | NASDAQ |
| Headquarters | Santa Clara, California |
| Incorporation | Delaware |
NVIDIA was founded in April 1993 by three engineers — Jensen Huang, Chris Malachowsky, and Curtis Priem — at a Denny's restaurant in San Jose. The company nearly went bankrupt in the late 1990s before a $5 million investment from Sega supported the RIVA 128 chip, and it has been continuously led by co-founder Jensen Huang for 33 years. The 1999 invention of the modern GPU set the foundation for the gaming market, and the 2006 launch of the CUDA software platform laid the foundation for the AI-compute market that now drives the company's results. NVDA went public in January 1999 at $12 per share; six stock splits later, that share would today represent the equivalent of an extraordinary multi-thousand-fold return.
This is the typical pattern with TGI rankings: the system surfaces companies whose numbers are exceptional regardless of whether you've heard of them. Brand recognition is not part of the scoring.
The price-growth figures compound across every timeframe, with the ten-year window in a different order of magnitude entirely — the signature of a genuine multi-decade compounder rather than a single-year momentum spike.
NVDA has paid a quarterly dividend since 2012, so no timeframe shows a 0% anomaly. The growth rates reflect a long sequence of modest year-over-year increases applied to a very small base dividend — the entire period predating the 2,400% increase announced after the close on May 20, 2026.
The four-year average yield reflects sustained payout strength over the trailing four years, during which NVDA's dividend was effectively a token. The 25x increase will shift the trailing yield over time, but even at the new annualized $1 per share the forward yield is only around 0.4% at recent prices — NVDA remains far outside the range the Healthy Income scoring system favors.
NVDA's dividend policy is discretionary, board-declared, and historically symbolic; the primary capital-return mechanism has been share repurchases. Before the June 2024 ten-for-one split the quarterly dividend was $0.04 per share; the split adjusted it to $0.01, where it stayed through Q1 FY27. On May 20, 2026, alongside the Q1 FY27 earnings release, the board declared a new quarterly dividend of $0.25 — a 2,400% increase — payable June 26, 2026 to holders of record June 4, 2026. Recent quarterly dividends ran $0.01 for each of Q2 FY26 through Q1 FY27, with the next payment stepping to $0.25.
At the new rate the annualized dividend produces roughly $24 billion of cash return per year against a share base of about 24.3 billion. The same May 20 release added $80 billion to the buyback authorization; combined with $62.2 billion remaining under prior programs, NVDA now has roughly $142 billion of buyback capacity — dwarfing the dividend by about six to one annually. The dividend hike is a maturation signal; the buyback remains the principal mechanism.
Total return over the trailing decade is approximately equal to price growth, since the dividend contribution across that window is negligible.
The underlying price-growth, dividend-growth, and yield figures are on the Stock Lookup.
NVDA designs graphics processing units (GPUs), accelerated-computing platforms, and the software stacks that bind them together. Starting Q1 FY27 the company reports two market platforms, Data Center and Edge Computing. Within Data Center it now breaks out two sub-markets: Hyperscale (public clouds and the largest consumer-internet companies) and ACIE (AI Clouds, Industrial, and Enterprise). Edge Computing covers PCs, game consoles, workstations, AI-RAN base stations, robotics, and automotive systems for agentic and physical AI.
By Q1 FY27 revenue, Data Center dominates: Hyperscale about $38.2 billion and ACIE about $37.0 billion, for a Data Center total of $75.2 billion (up 92% year over year), with everything else — gaming, pro visualization, automotive, and edge — about $6.4 billion, and total revenue of $81.6 billion (up 85%). Within Data Center, compute was roughly $60 billion (up 77% YoY) and networking roughly $15 billion (nearly tripling YoY). The number of partner data centers exceeding 10 megawatts has roughly doubled in the past year — an explicit indicator of the physical buildout behind the financial growth.
The compute-architecture roadmap runs from Hopper (H100/H200, transition largely complete) through Blackwell (B100, B200, GB200, roughly 70% of Data Center compute in the prior quarter) and the Blackwell 300 / GB300 Ultra ramp, to the announced Vera Rubin platform (Vera CPU plus Rubin GPU, shipments scheduled to ramp Q3 FY27), alongside BlueField-4 STX for accelerated storage in agentic AI factories and the open-source Dynamo 1.0 inference software now in production.
Fundamentals as of May 20, 2026.
NVDA reports both GAAP and non-GAAP EPS, with the gap most often driven by stock-based compensation. Beginning Q1 FY27 the company elected to include stock-based compensation in its non-GAAP measures, narrowing the historical gap. Q1 FY26 also carried a $4.5 billion charge tied to H20 inventory and purchase obligations after the April 2025 export-licensing changes, which made non-GAAP look notably better than GAAP for that single quarter. FY26 (ended January 25, 2026) GAAP net income was $120.07 billion, or $4.90 GAAP diluted EPS ($4.77 non-GAAP); Q1 FY27 was $2.39 GAAP and $1.87 non-GAAP diluted EPS.
Net cash from operations was $50.3 billion in Q1 FY27 alone, against $27.4 billion a year earlier, with quarterly free cash flow of $48.6 billion (versus $34.9 billion in Q4 FY26 and $26.1 billion in Q1 FY26). Full-year FY26 free cash flow was $96.5 billion, up from $60.7 billion in FY25 — the rare case at this scale where the year-over-year growth rate is itself accelerating.
The mechanism behind NVDA's rank is the interaction of Blackwell-driven Data Center revenue, software-platform gross margins, and capital-return policy. Roughly 75% hardware gross margins reflect the platform economics of CUDA lock-in rather than a typical hardware cost structure; those margins produce the free cash flow that funds the buyback program, which compresses the share count and, combined with growing net income, drives the EPS growth behind the price-growth component of the score. The May 2026 dividend hike does not change that mechanism so much as add a second channel: buybacks compress share count, the dividend distributes cash directly, and the framework reads the combination as a company moving past a growth-only phase.
NVDA carries essentially no net leverage — roughly $10.6 billion of cash against about $11.4 billion of total debt, an approximately neutral net position, rated A1/A+ investment grade — at a scale where almost unlimited debt would be available on demand. Choosing not to lever the balance sheet is itself a capital-allocation choice.
Net revenue grew 65% to $215.88 billion in FY26 from $130.50 billion in FY25, and Q1 FY27 revenue of $81.61 billion was up 85% year over year at roughly 75% gross margin. The quarter set records on revenue, free cash flow, and Data Center revenue; total revenue grew 20% sequentially and Data Center 21% sequentially. Management's Q2 FY27 outlook is $91 billion plus or minus 2% at 74.9%/75.0% gross margin, and explicitly excludes any Data Center compute revenue from China.
Recent analyst consensus places the twelve-month price target near $275.83, ranging from $140 to $380 across 61 covering analysts, with an average Strong Buy rating; post-earnings revisions on May 19–20 included HSBC $325, DA Davidson $300, Morgan Stanley $285, Wedbush $300, and KeyBanc $300. This is information, not endorsement: TGI scoring does not incorporate forward analyst targets, and consensus targets are more often wrong than right, in both directions.
The TGI scoring system surfaces companies based on past performance. It is not a prediction of future returns. NVIDIA Corporation carries several real risks that any reader should weigh independently of the scoring rank.
Customer concentration. Hyperscale customers — a handful of cloud providers and large consumer-internet companies — account for roughly half of Data Center revenue. Any combined slowdown in hyperscaler capital-expenditure plans, whether from a recession, an AI-cycle plateau, or a shift to in-house silicon, would hit NVDA's reported numbers immediately.
Sovereign exposure and export controls. NVDA's Q2 FY27 guidance explicitly assumes zero China Data Center compute revenue. Any further tightening of US export controls, or retaliatory action by other governments, removes geographic markets the company cannot replace quickly — and the reverse risk exists too: a sudden loosening could create inventory and pricing dislocations.
TSMC fabrication dependency. NVDA does not manufacture its own chips. The most advanced nodes in the Blackwell and Rubin platforms are fabricated by TSMC in Taiwan, with no alternative path to leading-edge fabrication at scale inside the planning horizon — which makes Taiwan Strait geopolitics an unhedgeable structural exposure.
Custom silicon competition. NVDA's largest customers all have incentives to reduce single-supplier dependency. Google's TPU, AWS's Trainium and Inferentia, Microsoft's Maia, Meta's MTIA, and Alibaba's newly announced AI chip are all efforts to internalize what NVDA provides. The CUDA software lock-in is a real moat, but the customers writing the largest checks are also the ones most able to invest in alternatives.
Valuation and expectations. NVDA has slid in after-hours trading following each of its last four earnings reports despite beating consensus each time — the bar is rising faster than the company can clear it on the headline numbers. A quarter that comes in merely in line rather than above could produce a sharper reaction than the fundamentals would justify.
Cyclicality of AI infrastructure spending. The roughly $725 billion of projected 2026 hyperscaler capital expenditure is not a permanent floor. AI infrastructure spending has historically moved in cycles, and this one has been long and intense; a digestion phase, if it arrives, would affect NVDA disproportionately given its position at the center of that spending.
The Coordination Geometry framework reads any organization across four abstract fields — Tribal, Jurisdictional, Economic, and Cultural — with four pillars — Capital, Information, Innovation, and Trust — doing the structural work inside those fields. NVIDIA Corporation is visible at all four field layers, and the pattern of which pillars stabilize which fields is more informative than any single financial metric.
Reading NVIDIA through this lens surfaces a structural pattern that the financial numbers alone obscure: a company whose Capital pillar is operating at full maturity, whose Innovation pillar has produced a durable cultural artifact in CUDA, whose Information pillar is in the middle of an honest disclosure transition, and whose Trust pillar carries both extraordinary strength (founder continuity, developer base) and structural fragility (sovereign permission gates, single-supplier fabrication). The fields show where NVDA is doing the work. The pillars show what is holding it up, and where the load-bearing elements have become single points of exposure.
The pillars doing structural work in NVDA's tribal field are Trust and Innovation. Thirty-three years of unbroken founder leadership is among the longest tenures in Silicon Valley history, and it represents a Trust pillar phenomenon in the framework's strict sense: an agreement between founder and institution that has been validated quarter after quarter for three decades, producing the kind of commitment no governance structure can manufacture in less time. The CUDA developer ecosystem of more than five million practitioners is a second Trust artifact, a community held together not by employment contracts but by the validated reliability of the toolchain and the durable agreement that NVIDIA will keep the platform stable and forward-compatible. This is the Innovation pillar's output, idea times experimentation producing solutions, crystallized into a tribal identity that operates entirely outside the company's payroll.
The cultural literacy of recent product naming (Vera, Rubin) is itself a tribal signal aimed at a technically informed adopter base. Naming a platform after the astronomer who discovered dark matter is a gesture only legible to a particular audience, and it functions the way uniform insignia function in a regiment: identification of the in-group. Massive stock-based compensation has converted much of the workforce into quasi-owners, which moves the employee relationship from a Capital exchange (wages for labor) toward a Trust position (shared stake in validated outcomes). The 10-for-1 split in June 2024 was explicitly framed around employee and retail accessibility, which reads structurally as widening the Trust base rather than optimizing for any particular shareholder class.
The honest tension to name is that thirty-three years of one CEO produces institutional memory but also concentrates judgment. The leather jacket has become a brand asset, and Huang's personal presence at the President's Council on Science and Technology, on Air Force One to China, at every keynote, is a coordination shape with a single load-bearing node. Strong Trust pillars often look this way before they are tested by succession, and NVDA has not yet been tested.
The pillars active in NVDA's jurisdictional field are Trust and Information, with Capital underwriting both. The structural pattern is that NVDA's products have become objects of sovereign concern at a speed exceeding the framework any single jurisdiction has built around them. The company is now guiding Q2 FY27 revenue with zero assumed China data center compute contribution, which absorbs a binary regulatory outcome into the financial baseline as if permanent. At the same time, sovereign AI as a customer category produced over $30 billion in FY26 revenue and more than tripled year over year. Sovereigns increasingly sit on both sides of NVDA's customer ledger, as buyers in some jurisdictions and as gatekeepers in others.
Huang's appointment to the President's Council of Advisors on Science and Technology, and the Air Force One trip alongside Trump and Musk in May 2026, are not personal anecdotes but structural facts. The company has embedded its leadership directly into US sovereign decision infrastructure because the alternative is being acted upon rather than consulted. The new Hyperscale and ACIE segmentation in the Data Center business is an Information pillar move that separates the jurisdictionally concentrated hyperscaler customers from the more geographically distributed sovereign and enterprise base. The reporting structure is itself a legible reading of which buyers face which regulatory pressures.
The unhedgeable exposure is TSMC. No amount of design innovation closes the gap between NVDA's intellectual property and the Taiwan Strait. This is a Trust pillar dependency on a single counterparty in a single jurisdiction, and the framework cannot read it as anything other than load-bearing fragility. The company knows this, which is why the Arizona fab investments and broader diversification are underway, but the timeline of fabrication relocation is longer than the timeline of geopolitical risk evolution. The company has bought time, not closure.
NVDA's economic field is the cleanest example on the watchlist of a Capital pillar functioning at full operational maturity. The framework defines Capital as Stock times Velocity producing Work. NVDA's stock position is extraordinary at $5.4 trillion in market capitalization with no net leverage, 75% gross margins, and $96.5 billion in free cash flow. Its velocity is unmatched in the public markets. The work being produced is the physical buildout of AI infrastructure at hyperscale. The capital structure is wealth-based by the framework's strict definition: returns flow from verified present positions in cash and free cash flow, not from debt extracted against imagined futures. The company carries essentially zero net debt at a scale where leverage would be available on any terms it asked for.
Today's dividend raise from $0.01 to $0.25 per share is a 2,400% increase that nonetheless remains symbolic against the buyback architecture. An annualized $1 per share dividend produces roughly $24 billion of return per year. The buyback authorization now stands at approximately $142 billion. The dividend is the maturation signal, communicating that NVDA now has the cash flow visibility to commit to a per-share distribution. The buyback remains the primary capital return mechanism. This is a Capital pillar reading the difference between a commitment that scales with share count (dividend) and a commitment that compresses share count (buyback). Both are valid wealth-based instruments. Together they signal a company that has moved past the growth-only phase into one where capital return discipline is being explicitly demonstrated.
The decision to include stock-based compensation in non-GAAP measures beginning Q1 FY27 is an Information pillar maturation. The company is no longer asking investors to look past employee equity dilution. The validation chain, data times verification producing proof, is being shortened, and the implicit Trust agreement with the investor base is being tightened. The honest tension to name is that this extraordinary Capital position rests on a customer base concentrated enough to resemble a defense procurement program. Hyperscale alone produces more than half of Data Center revenue. The Capital pillar looks wealth-based at the company level. The revenue dependency looks more concentrated than the framework prefers.
NVDA's cultural positioning is archetypal infrastructure. The pillars doing the work are Innovation (CUDA as cultural artifact), Information (the narrative is itself an information output that shapes market and policy behavior), and Trust (the "AI factory" framing implies durable agreement structures that will hold beyond any single quarter). Huang's repeated framing of AI as the next electricity, the next internet, positions NVDA archetypally not as a competitor inside a market but as a substrate beneath one. "It runs on CUDA" now carries the cultural weight that "it runs on Windows" carried in the 1990s, and the analogy is structurally accurate: a proprietary platform that has become close enough to a standard that the developer base is locked in by its own training and its own toolchains, not by contract.
The product naming reinforces the substrate positioning. Vera Rubin discovered dark matter, the invisible mass that holds galaxies together, and the naming gesture casts NVDA in the same role for computation: making visible what is currently invisible, providing the substrate that the work depends on. "Agentic AI has arrived" is a sovereignty-of-narrative move, the kind a company makes when it intends to define the next paradigm rather than respond to it. The "AI factories" metaphor deliberately makes computation physical, capital-intensive, and infrastructural rather than abstract, which is a cultural choice with structural consequences: it justifies the scale of customer commitment and the depth of the buildout that customers are now committing to.
The honest tension is sharpest here. NVDA's narrative is democratization and ubiquity, "every cloud, every industry," AI for everyone. The customer book reads differently. Hyperscale is more than half of Data Center revenue, sovereign AI is the next concentration vector, and the export control regime makes clear that NVDA's chips are not in fact like electricity. Electricity is not subject to sanction. NVDA's products operate inside sovereign permission gates, and the cultural framing as ubiquitous infrastructure coexists with a customer concentration that more closely resembles a Cold War defense contractor than a utility. The framework does not call this a failure. It calls it the shape of the coordination system as it actually is: a substrate aspiration operating through a small number of very large nodes, dependent on a single fabrication partner, embedded in a single sovereign's strategic apparatus, holding a developer base that genuinely is distributed. All four readings are simultaneously true. The coordination shape is the tension between them.
Sector cap: Information Technology is a standard GICS sector, governed by the 10% per-sector cap; when a portfolio's IT exposure is at or above that cap, no new direct NVDA buys can be made even though the individual rank sits high in the buy zone.
Position cap: The standard 5% per-position cap applies. NVDA is an ordinary operating company, not a leveraged product, so no tighter structural cap is triggered — though in a portfolio already concentrated in a few mega-cap technology names, adding NVDA reinforces an exposure that may already be dominant.
Account placement: US-domiciled (Delaware), so no foreign dividend-tax-credit consideration; the dividend is small enough that income placement is not a meaningful driver either way.
Dividend concentration: NVDA's dividend is token even after the May 2026 hike (a forward yield near 0.4%), so it contributes almost nothing to the 5% dividend-concentration cap and is not an income holding under any strategy.
Notes: This is the rare case where a household name and a top-of-list rank align, and the discipline still governs the action: the ranking surfaces the opportunity, and the sector cap can deny it. Indirect NVDA exposure keeps growing through semiconductor-industry ETFs and other IT positions that hold it, so a capped portfolio is not cut off from the trajectory — it just is not adding direct shares. The rank is the rank for everyone; buyability depends on portfolio shape, which is each reader's own.
This is the discipline the framework is built around. The scoring system surfaces opportunities; the diversification rules govern action. When the two conflict — and they will, regularly — the rules win. Your own portfolio's caps and holdings will differ, so treat these as the rule, not a recommendation.
Discipline builds, speculation crashes.Total Growth Investing