I'll pull up NVIDIA's gross profit, total revenue, and operating income data to calculate both margins.Now I'll compute the annual gross margin and operating margin and create the chart.

Loading chart...Here are NVIDIA's annual gross and operating margins over the past 5 fiscal years:

The chart highlights a sharp margin compression in FY23 during the post-crypto inventory correction, followed by a dramatic recovery as the AI/data center boom took hold. FY25 marked the peak at 75% gross and 62% operating margins, with a slight moderation in FY26 as the business scaled further.

Loading chart...Here are NVIDIA's gross and operating margins over the last 10 fiscal years:

The 10-year view shows three clear phases: steady margin expansion from FY16–FY22 as NVIDIA transitioned from PC graphics into data center and AI, the sharp FY23 correction during the post-crypto inventory glut that pushed operating margin to just 15.7%, and the explosive AI-driven recovery from FY24 onward that took both metrics to all-time highs before a modest pullback in FY26.

Great question. The answer lies in operating leverage — when revenue scales rapidly, fixed operating costs (R&D, SG&A) grow much slower than gross profit, so each incremental dollar of gross profit flows disproportionately to operating income.

Let me pull the actual numbers to quantify this.Here's the key insight — the numbers tell the story clearly:

The Operating Leverage Effect

The three drivers:

1. Gross margin expansion (+10pp): A real contributor, driven by the shift to high-margin data center products (H100, Blackwell) and pricing power. But it's only part of the story.

2. Massive operating leverage from OpEx: This is the bigger force. Operating expenses as a percentage of revenue collapsed from ~36% in FY16 to just 12.6% in FY25. R&D fell from 27% to 9.9% of revenue, and SG&A from 12% to 2.7%. NVIDIA's revenue surged nearly 5× from FY22 to FY25 while R&D and SG&A grew far more slowly in dollar terms — they're largely fixed costs.

3. The math: When gross margin rises 10pp and OpEx ratio falls 15pp, operating margin expands by roughly 25pp — exactly what we see. The operating margin swing (37% → 62%) is simply the sum of both tailwinds compounding.

FY23 is the mirror image: revenue stalled at $27B (flat vs FY22), gross margin fell to 57%, yet OpEx stayed elevated at 36% of revenue. The result: operating margin collapsed to 15.7%. Same operating leverage, working in reverse.

Good question — let's model this. We need FY26 data (the most recent full year) and then apply the pre-AI-boom gross margin while keeping revenue and OpEx dollars intact.Now let me compute the scenario. I'll use FY26 as the base and apply the pre-AI gross margin (~62%, the average of FY16–FY22).Here's the modeled impact — the headline is that operating income is surprisingly resilient:

The key takeaway

Even if gross margin collapsed all the way back to FY23's 57% — while holding today's revenue and OpEx dollars constant — NVIDIA would still deliver $100 billion in operating income at a 46.3% operating margin. That's roughly flat with FY25's actual operating income of $81.5B.

The reason: OpEx is now just 10.7% of revenue. In FY22, OpEx was 27.6% of revenue. So that ~17 percentage points of structural OpEx leverage cushions the blow. A 14pp gross margin drop only translates to a 14pp operating margin drop because there's far less OpEx to amplify the pain anymore.

The more realistic pre-AI scenario (62% GM) would only cut operating income by 15% — to ~$111B — still a massive number and a 51% operating margin. The operating leverage that supercharged the upside now serves as a powerful floor on the downside.

Not quite — in fact, the opposite. Let me pull apart the two effects to show what's really happening.Let me now isolate the key transition years to make the point crisply.No — the data actually points the other way. Here's the story:

Pricing power, not just volume, drove the margin expansion. When gross margin jumps from 57% to 75%, it means revenue per unit is growing faster than cost per unit. If this were purely a volume story at stable prices, gross margin would stay flat.

Look at FY23 → FY24, the AI breakout year: revenue surged +126% while COGS rose only +43%. That gap — revenue growing nearly 3× faster than input costs — is the signature of massive pricing power. NVIDIA wasn't just selling more chips; they were commanding dramatically higher prices per unit relative to what it cost to manufacture them.

What's actually happening:

• Volume surged massively — data center GPU shipments went exponential
• Pricing also jumped dramatically — H100/B200 systems sell for far higher ASPs than the gaming/data center mix of the pre-AI era, and at much richer markups over manufacturing cost
• The combination produced the 71–75% gross margins we see

The FY25 → FY26 data is interesting though: revenue grew 65% but COGS grew 91%. That's the first sign of pricing power easing — costs are catching up, likely as Blackwell ramps (new architectures start with lower yields) and competition/mix shifts. That's why gross margin ticked down from 75% to 71%.

So the answer is: both volume and pricing surged together, but pricing power (ASP growth outstripping unit cost growth) was the dominant driver of the margin expansion. The earlier model I ran — where gross margin falls back to 62% — is effectively modeling what happens if that pricing premium evaporates while volume stays elevated.

We already have this from the earlier model, but let me isolate it cleanly with the exact pre-AI pricing assumption. Let me also back out the implied ASP vs cost dynamics to quantify the "pricing premium" explicitly.Here's the quantified answer:

NVIDIA FY26: If AI Pricing Power Vanishes

The punchline

Gross profit drops from $153.5B → $131.2B. Operating income drops from $130.4B → $108.1B.

That's a 17% cut to operating income — meaningful, but not catastrophic. Operating margin still lands at 50%, and $108 billion of operating income would still be the second-best year in the company's history (behind only FY26 itself).

The reason the damage is contained: the 10.3 percentage points of gross margin that came from AI-era pricing power flow almost dollar-for-dollar to operating income, because OpEx is already so lean at just $23B. There's no double-counting — every dollar of lost gross profit is a dollar of lost operating income. But with $216B of revenue, a 10pp gross margin swing is a $22B hit against a $130B base, not a wipeout.

Excellent point — yes, in reality, a pricing decline would almost certainly mean some revenue decline too. Volume wouldn't stay perfectly intact if NVIDIA had to cut prices back to pre-AI levels; that implies a much more competitive market. Let me model a few realistic scenarios that combine pricing and volume effects.You're right — and once you layer in the revenue hit, the picture changes materially:

The real risk: OpEx turns from tailwind to headwind

The $23B OpEx is the fixed anchor. In FY26, it's just 10.7% of revenue — a rounding error against $153B of gross profit. But as revenue shrinks from both lower pricing and lower volume, OpEx becomes an increasingly heavy burden:

• At $183B revenue (mild), OpEx is still only 12.6% — manageable, op income down 30%
• At $151B (moderate), OpEx hits 15.3% — operating income halved
• At $108B (severe), OpEx is 21.4% — operating income down 70% to $38.5B

In the moderate scenario, operating income falls back to roughly FY24 levels ($67.6B vs $33B actual FY24), still a very large profit pool. But the severe case — a genuine pricing war with 50% revenue destruction — would wipe out $92B of operating income. The operating leverage that was a superpower on the way up becomes a vice on the way down once revenue shrinks enough for OpEx to matter again.

Great question — let me decompose it properly. We need to separate revenue growth into a volume component and a price component. The key insight: changes in COGS per unit tell us about unit costs, and changes in gross margin tell us about pricing relative to cost.Here's the decomposition. Let me walk through it.

The Markup Ratio: Revenue per Dollar of Cost

This is the cleanest way to separate volume/cost scale from pricing power:

FY22 → FY26: The $189B Revenue Jump Decomposed

Revenue grew from $26.9B to $215.9B — a $189B increase:

The answer

80% of the revenue surge was volume-driven — NVIDIA shipped dramatically more units. But that understates pricing, because the 17% "interaction" term is where pricing power multiplies against that huge volume. The markup ratio went from 2.85× (FY22) to a peak of 4.00× (FY25) before settling at 3.46× in FY26.

In concrete terms: for every dollar of manufacturing cost, NVIDIA used to generate ~$2.60–2.85 of revenue in the pre-AI era. During the AI boom, that jumped to $3.67–4.00. That's a pricing premium of roughly 30–40% over pre-AI markup levels. On $215B of revenue, that premium alone is worth tens of billions.

So the story is: volume exploded (80% of the revenue delta), but pricing power amplified that volume by 30–40%, and that amplification — applied against an enormous volume base — is what created the extraordinary margin structure. Price didn't double; the markup per dollar of cost rose ~35%. But at $216B of revenue, that 35% markup premium translates to ~$50B+ of incremental gross profit.

Let me model this precisely: strip out the 35% pricing premium from FY26, keep volume (COGS) constant, and see what happens to revenue, gross profit, and operating income.Here's the clean answer:

FY26: Remove the 21% Pricing Premium, Volume Unchanged

The bottom line

Stripping out just the pricing premium — keeping the same units, same COGS, same OpEx — revenue drops by $38B (18%) and operating income drops by $38B (29%), landing at $92.5B.

That's still a phenomenal business: $178B revenue, 65% gross margin, 52% operating margin, and nearly $93B of operating income — numbers no semiconductor company has ever come close to. But the pricing premium alone was worth $38 billion in operating profit, which is more than NVIDIA's entire operating income in any year before FY24.