NVIDIA Opens The Rack, Keeps The Rent
Updated: Oct 15
Why NVIDIA deserves attention today
A blueprint handed to the industry
On 12 October 2026, at the OCP Global Summit in San Jose, NVIDIA contributed key Vera Rubin rack and compute tray designs to the Open Compute Project through its open MGX architecture. Cisco, Google, Meta, Microsoft and Visionbay are among the companies advancing with Vera Rubin. The same day brought open standards for AI factory design and grid resilience, new open source software for NVIDIA DSX AI factories, and networking advances with Meta, Cisco and Google.
Generosity, or capital allocation?
When a market leader publishes the designs behind its flagship systems, the first reading is generosity. A board reads it differently. Open standards let equipment makers, colocation providers and data center operators design and build AI factory infrastructure without waiting for proprietary specifications from individual customers. In management accounting terms, the ecosystem shares the cost, the capital and the time required to build out demand for NVIDIA's platform.
The scorecard behind the press release
Press releases describe intentions; ratings measure results. WorkN'Play's Corporate Intelligence App rates NVIDIA 68.79 out of 100, the highest score and the only "Very High" overall rating in its Artificial Intelligence panel, 7.49 points ahead of the runner-up. The ratings are comparative: they position each company against its panel and weight momentum above size.
Why momentum changes the reading
The App weights the change in each indicator more heavily than its current level. A strong snapshot with a deteriorating trend can score below a modest snapshot that is improving. NVIDIA's lead therefore says more than "large and profitable": the company is still strengthening on most fronts. The same method also shows where the trend works against it, and that is exactly where the OCP announcements become most revealing.
The overall picture: NVIDIA against the industry
Eight "Very High" ratings out of twelve
NVIDIA scores "Very High" on 8 of its 12 sub-ratings and ranks first in the panel on six of them:
human capital
production assets (shared)
research and development
profitability
total shareholder return
economic value added
It ranks second in the panel on bargaining power and corporate debt.
A complete value-creation chain
The strongest indices form a chain, each link feeding the next:
Growth: revenue compounds at 100.0% a year against 89.4% for the industry.
Margin: the operating margin reaches 60.4% against 57.9%.
Returns: average return on total assets is 56.2% against 44.2%.
Value created: cumulative economic value added is $149.0 billion against $23.5 billion.
Shareholder reward: cumulative shareholder return is 368.2% against 165.8%.
Efficiency that scales with growth
This growth is not bought with overheads:
Revenue per employee is $5.14 million.
Every dollar of R&D returns $16.70 of revenue.
Asset efficiency exceeds 100%.
Customers pay NVIDIA before NVIDIA pays its suppliers.
MSGA expenses take only 5.4% of total expenses, against 9.4% for the industry.
Three amber lights
Three indices call for board attention:
Cost of goods sold management (37.04): last in the panel.
ESG risk management (57.41): last in the panel.
Working capital management (39.58, Medium Lower): fourth of six.
A common thread: trajectory, not level
In each of the three, the issue is momentum:
The cost mix is tilting further toward materials.
Environmental and social risk indices are rising faster than the industry's.
Working capital metrics only track the industry rather than beat it.
One further watch point sits inside a strong index: receivables lengthen faster than the industry's.
The twelve dimensions in detail
1. Human capital: $5.1 million per head, and rising
Index: 75.00, Very High, first in the panel
Output per person in a league of its own
Revenue per employee reaches $5.14 million, 43% above the industry's $3.60 million. It is also growing faster: 70.9% a year against 67.5%. NVIDIA leads on both the level and the trend, which is the combination a momentum-based model rewards most.
Hiring fast, but revenue runs faster
NVIDIA is expanding its workforce faster than the industry, with headcount up 17.0% a year against 13.1%. Yet payroll takes a smaller share of total expenses (13.8% against 16.4%), and that share is falling faster (−19.2% against −17.7%). The reason is simple: revenue grows at 100.0% a year, far ahead of headcount.
The board's reading
This is operating leverage on talent. Every new hire is absorbed by a revenue base that grows roughly six times faster than the workforce. The strategic question is how long this ratio can be sustained as NVIDIA broadens from chips into software layers such as DSX OS.
2. Bargaining power: paid before it pays
Index: 81.67, Very High, second in the panel
Suppliers fund the receivables
NVIDIA collects from customers in 65 days and pays its suppliers in 71. Its suppliers therefore finance its receivables, with a six-day cushion against the industry's three (66 and 69 days). For a company of NVIDIA's size, that cushion represents substantial free financing.
A cycle shrinking faster
The cash conversion cycle is shrinking at an average annual rate of 16.9%, against 11.8% for the industry. Collection and payment terms are both moving the other way, so the improvement comes from inventory, where days outstanding fall 23.3% a year.
The factoring lens
Specialists in receivables financing will spot the caveat:
Days sales outstanding are lengthening by 7.9% a year, 1.8 times the industry's 4.4%.
Days payable are shortening by 2.9% a year, against 2.6% for the industry.
Payables are growing more slowly than receivables (ratio 0.9 against 1.0).
If these trends persist, the cushion erodes. As customers commit to gigawatt-scale AI factories, credit terms deserve as much scrutiny as order books.
3. Cost of goods sold: the bill of materials bites back
Index: 37.04, Very Low, last in the panel
Hardware gravity
Cost of revenues takes 73.0% of NVIDIA's total expenses, against 68.2% for the industry. Costs of materials consumed take 59.2% against 55.8%, and that share is rising faster: 33.6% a year against 30.4%. In a model that rewards momentum, a cost structure growing heavier faster than the industry's is penalised.
Margin is still widening
This is not a margin crisis. Cost of revenues grows at 75.2% a year, below revenue growth of 100.0%, so gross margin keeps expanding. The index flags the direction of the cost mix, not current profitability. Boards should treat it as an early warning rather than an alarm.
Inventory: heavier, but leaner each year
NVIDIA holds 113 days of inventory against 110 for the industry. Inventory days are falling faster, though, at 23.3% a year against 17.0%. This is also what drives the faster shrinking of the cash conversion cycle.
Where the OCP agenda helps
Rack-scale systems are material-intensive. MGX 1.2, for example, adds a liquid-cooled 5,000A busbar and rack-scale blind-mate liquid cooling. The specification is also designed to improve assembly, test and service times, and it standardises power and cooling interfaces. Open designs let the ecosystem share the manufacturing load, which points in the right direction for this index.
4. Production assets: sweating the assets above 100%
Index: 64.81, Very High, tied for first in the panel
Assets working beyond full rate
Asset efficiency stands at 104.4%, against 95.3% for the industry. The productive asset investment ratio is 1.5 against 1.2, which signals heavier reinvestment in the productive base. NVIDIA is both using its assets harder and renewing them faster.
Capex productivity
Each dollar of capex generates $35.74 of revenue, slightly below the industry's $36.56. This ratio is improving faster, however: 0.34 against 0.31.
The momentum caveat
Two indicators are growing more slowly than the industry's:
asset efficiency: 16.8% a year against 24.6%
productive asset investment ratio: 21.4% against 24.7%
NVIDIA's lead rests on levels more than on pace here.
The ecosystem as capacity
MGX 1.2 supports rack deployments up to 250kW. The open specifications give manufacturers worldwide a common target to build to. Capacity that extends NVIDIA's reach is increasingly built on its partners' balance sheets rather than its own, a structurally capital-light way to scale.
5. MSGA expenses: lean overheads, loud ecosystem
Index: 61.67, Very High, fourth in a closely bunched panel
Lean by design
MSGA takes only 5.4% of NVIDIA's total expenses, against 9.4% for the industry. Few companies of this scale carry such a light commercial and administrative structure.
The industry is trimming faster
The industry is reducing its MSGA share slightly faster than NVIDIA: −21.8% a year against −20.7%. NVIDIA starts from a much lower base, so each further point of reduction is harder to achieve.
Returns on overheads
Return on MSGA grows at 62.2% a year, a remarkable pace in absolute terms. The industry grows it slightly faster, at 64.6%.
Why fourth despite a superior level
In a momentum-driven model, a small lag on both trends outweighs a large advantage in level. That explains fourth place in a panel where the top four companies are separated by only 3.33 points.
Partners as a sales force
The OCP announcements show the commercial model at work:
Meta is adopting Spectrum-6 switches.
Cisco is co-developing the USFP optical module, with TE Connectivity and Amphenol joining the ecosystem.
Google is bringing its Falcon transport protocol to future ConnectX-10 network cards.
Each partner that builds on NVIDIA's architecture also sells it, extending NVIDIA's commercial reach without a proportional rise in selling costs.
6. Research and development: sixteen dollars from every lab dollar
Index: 72.22, Very High, first in the panel
Superior returns on research
Each dollar of R&D returns $16.70 of revenue (industry: $14.60) and $11.90 of gross profit (industry: $10.40).
Returns that compound
Both returns are growing faster than the industry's:
revenue per R&D dollar: 48.4% a year against 43.6%
gross profit per R&D dollar: 66.4% against 56.8%
The lab is becoming more productive each year, not less.
A falling share is not a retreat
R&D takes 21.6% of total expenses (industry: 22.4%), and that share is falling by 12.5% a year (industry: −10.0%). The falling share reflects a cost base expanding with volumes. The rising returns confirm that innovation is not being sacrificed.
Research turned into standards
The OCP announcements show R&D becoming industry infrastructure:
Spectrum-6 delivers twice the switching capacity of Spectrum-4.
Spectrum-X Multiplane spreads GPU traffic across up to eight network planes, letting AI factories scale beyond 100,000 GPUs.
USFP offers four times the bandwidth density of traditional 200Gb/s-per-lane pluggable optics.
Standards are the most durable return a lab can earn.
7. Working capital: comfort is not a strategy
Index: 39.58, Medium Lower, fourth in the panel
Comfortable liquidity
NVIDIA's working capital ratio stands at 3.9 (industry: 3.8) and has averaged 4.2 (industry: 4.0). Short-term obligations are covered nearly four times over.
Lean working capital per dollar of revenue
NVIDIA ties up 0.43 of working capital per dollar of revenue, against 0.46 for the industry, or 0.49 against 0.53 on average. By this measure, it runs its working capital more efficiently than its peers.
Parity is not momentum
The trend, however, is almost identical to the industry's: the working capital-to-revenues ratio changes by −0.1 a year for both. A model that rewards relative momentum rarely rewards parity.
The board's question
The liquidity ratio is edging up, with average annual growth of 0.8 against 0.7. Is this buffer strategic, or simply idle? The lengthening collection period noted under bargaining power will absorb more working capital over time, so receivables management is the lever to watch.
8. Profitability: ninety-six out of a hundred
Index: 96.30, Very High, first in the panel
Close to a perfect score
This is NVIDIA's highest sub-rating and the strongest pillar of its overall rating.
Gross margin in line, rising faster
Gross margin stands at 71.1%, in line with the industry's 71.3%. It is expanding faster, though: 8.5% a year against 6.8%.
Operating leverage below the gross line
The advantage builds further down the income statement:
operating margin: 60.4% against 57.9%
net margin: 55.6% against 53.5%
Lean overheads and a falling R&D share turn gross profit into operating profit more efficiently than the industry does.
Reading the growth rates correctly
The industry shows faster margin growth: 131.0% against 85.9% for operating margin, and 119.7% against 71.9% for net margin. This typically reflects companies improving from lower starting margins. NVIDIA is compounding from the top of the range.
9. Corporate debt: a net-cash fortress
Index: 64.81, High, second in the panel
Equity-funded growth
NVIDIA's debt-to-equity ratio of 0.3 matches the industry's. Its leverage rate of 131.5% is close to the industry's 130.3%. Growth is financed overwhelmingly by equity and internally generated cash.
Deleveraging faster
NVIDIA is reducing its debt faster than the industry:
debt-to-equity: −28.6% a year against −16.3%
leverage rate: −11.0% a year against −4.9%
More cash than debt
A net debt-to-EBITDA ratio of −0.4 (industry: −0.5) means cash exceeds debt. This position has held broadly stable (−0.6% a year against +29.2% for the industry).
Why it matters strategically
A company inviting the world to build on its architecture needs a balance sheet nobody questions. Partners committing to multi-year AI factory roadmaps are, in effect, underwriting NVIDIA's continuity, and a net-cash position gives them that assurance.
10. Total shareholder return: 368% returned, dividends accelerating
Index: 83.33, Very High, first in the panel
More than double the industry
Cumulative shareholder return reaches 368.2%, against 165.8% for the industry. Over three years, the share price has compounded at 67.3% a year, against 38.5%.
Dividends growing three times faster
Dividend per share has grown at 36.9% a year, against 10.5% for the industry. Dividend growth trails share price growth, so investors are rewarded mainly through capital appreciation.
Returns on equity
Return on equity stands at 76.3% (industry: 66.4%) and has averaged 79.2% over the period (industry: 59.8%). The industry's faster ROE growth (183.2% against 88.7%) comes from a lower base.
11. Economic value added: clearing a 17.7% hurdle at scale
Index: 91.67, Very High, first in the panel
A demanding cost of capital
NVIDIA's weighted average cost of capital is 17.7%, more than two and a half times the industry's 6.8%. That premium is consistent with a high-growth, equity-financed profile. It is rising more slowly than the industry's, however: 50.9% a year against 83.9%.
Returns that clear the hurdle
Average return on total assets stands at 56.2%, against 44.2% for the industry, comfortably above NVIDIA's cost of capital. The industry is growing this return faster (201.7% against 120.0%), again from a lower base.
Value created at a different scale
Cumulative economic value added reaches $148,990 million, 6.3 times the industry average of $23,470 million. The industry grows its economic value added faster (1,156.0% against 658.8% a year), but what creates wealth for shareholders is the absolute value added.
The board's reading
A 17.7% cost of capital imposes discipline: every new investment must clear a high bar. Asking the ecosystem to share the capital burden of AI factories is consistent with that discipline.
12. ESG risk: the watt becomes the risk unit
Index: 57.41, Very Low, last in the panel
A reassuring snapshot
NVIDIA's three risk indices are all below the industry's 16.7%:
environmental: 15.3%
social: 9.7%
governance: 16.1%
A score of 57.41 ranks last only because every peer scores above 70.
A worrying trajectory
The trend tells a different story:
Environmental risk: rising 14.6% a year, against 2.6% for the industry.
Social risk: rising 7.4% a year, against 0.7%.
Governance risk: falling 0.3% a year, while the industry's edges up 0.2%; this is the only improving dimension.
Where OCP responds
Several announcements target the energy footprint of AI factories growing from megawatts to gigawatts:
NVIDIA's 800 VDC power architecture, adopted in Visionbay's K-1 design with Vertiv.
A grid resilience specification under which AI factories restore near-full load in less than two seconds.
USFP optics with roughly 1.3x greater power efficiency.
What remains unaddressed
The OCP agenda speaks to environmental risk. The faster rise in social risk is not covered by these announcements and deserves a separate place on the board's agenda.
Conclusion: reading direction, not just position
A value engine with three amber lights
NVIDIA emerges as the panel's clearest value creator: superior growth, near-perfect profitability, a net-cash balance sheet and economic value added six times the industry average. Three areas need watching: a cost mix tilting toward materials, working capital that only tracks the industry, and environmental and social risks rising faster than its peers'.
From headlines to evidence
Read alone, the OCP announcements are a technology story. Read against momentum-based ratings, they look like a coherent response to the weakest parts of NVIDIA's own scorecard. Open designs spread manufacturing load and capital across the ecosystem, and power-efficiency standards address the steepest risk trend.
Momentum as an early warning system
Static ratios describe where a company stands. Momentum shows where it is heading, often before the financial statements make it obvious. The NVIDIA case shows both sides: momentum confirms its lead on value creation and flags its cost and ESG trajectories while they are still manageable.
The value of the Corporate Intelligence App
This analysis was made possible by the WorkN'Play Corporate Intelligence App, developed by Jean Jacques André. The App:
benchmarks more than 400 corporations across 50 industries;
runs half a million calculations over more than 70 management accounting and financial indicators;
condenses them into 12 performance indices and one overall rating.
The boardroom takeaway
For directors, investors and lenders, the App turns a mass of data into the questions that matter: is this company getting stronger, where exactly, and what should the board watch next?
---------------------------------------------------------------------------------------------------
Analysis powered by the WorkN'Play Corporate Intelligence App, developed by Jean Jacques André, Founder & CEO of WorkN'Play, Chairman of the Board of MauFactoring Ltd., and Board Director of MauBank Holdings Ltd.
