Battle Stocks: NVIDIA vs. AMD

Welcome to Battle Stocks

This is the first installment of Battle Stocks, a new weekly series from Wealth Engine Pro Insights. The concept is simple: take two or three companies competing in the same market, put their numbers side by side, and let the data pick a winner. No cheerleading. No tribal allegiances. Just financials, fundamentals, and an honest assessment of which company deserves your capital.

Every Battle Stocks article will follow the same format: a head-to-head comparison of the key metrics, a deep dive into each company's strengths and weaknesses, a transparent discussion of the risks on both sides, and a clear verdict based on the data. We will also be asking you, our readers, to tell us which battles you want to see. If there are two companies you have been debating in your own portfolio, let us know. The best suggestions will become future installments.

We are starting with the biggest matchup in the market: NVIDIA (NVDA) versus AMD (AMD). The undisputed king of AI chips against the challenger that just landed the largest GPU deal in history. Let the data fight.

The Tale of the Tape

The raw numbers tell you two things immediately. First, NVIDIA is not just bigger than AMD. It is operating in a different financial universe. Its annual revenue is 6.2 times larger. Its net income is 28 times larger. Its gross margin is 21 percentage points wider. If this were a boxing match, NVIDIA would be the heavyweight champion and AMD would be the welterweight who just won a title fight and is moving up in class.

Second, AMD's stock has outperformed NVIDIA's by a factor of more than three over the past year. That kind of price action tells you the market is not pricing what AMD is today. It is pricing what AMD could become. The question is whether that future is priced correctly.

The NVIDIA Case

NVIDIA's financial profile in fiscal year 2026 (ending January 2026) is arguably the most impressive single-year performance by any technology company in history.

Revenue rose every quarter: $44.1 billion in Q1, $46.7 billion in Q2, $57 billion in Q3, and $68.1 billion in Q4. Full-year revenue hit $215.9 billion, up 65% from the prior year. Data center revenue alone was $197.3 billion, up from $115.2 billion the year before. Net income for the year was $120.1 billion. Operating income more than doubled to $130.4 billion.

Read those numbers again. NVIDIA generated more net income in a single year than AMD has generated in its entire 55-year history as a public company. That is not an exaggeration. It is arithmetic.

Q4 was the exclamation point. Revenue of $68.1 billion beat estimates of $66.1 billion. Data center revenue hit $62.3 billion, up 75% year over year. EPS of $1.62 beat the estimate of $1.54. And management guided Q1 FY2027 to $78 billion in revenue, plus or minus 2%, which represents yet another sequential step up.

The margin story is equally striking. NVIDIA's gross margins have hovered in the 73-75% range, a level typically associated with monopoly-grade software companies, not hardware manufacturers. The company's return on equity is 101.5%. Its debt-to- equity ratio is 0.07, meaning it is essentially debt-free. At a trailing P/E of roughly 40x and a forward P/E of 24x, with a PEG ratio of 0.61, NVIDIA is arguably not expensive relative to its growth rate. That is a counterintuitive statement about a $4.8 trillion company, but the math supports it.

NVIDIA reports its next earnings on May 20, 2026. The market expects $78 billion in revenue for the quarter. If they deliver, the annualized run rate will exceed $300 billion.

The AMD Case

AMD's story is not about being bigger than NVIDIA. It is about being the only credible alternative in a market that desperately wants one.

Fiscal year 2025 was a breakout. Revenue hit $34.6 billion, up 34% year over year. Data Center revenue reached $16.6 billion, driven by EPYC server CPU share gains and the ramp of Instinct MI300 and MI325 AI accelerators across every major hyperscaler: Meta, Microsoft, Google, and Amazon. Net income of $4.3 billion represented a 164% increase over the prior year. Q4 revenue was a record $10.3 billion, with Data Center at $5.4 billion. Non-GAAP EPS for the quarter was $1.53, up 40%.

But the number that changed everything is not in the earnings report. It is in the press release from February 24, 2026.

AMD and Meta announced a multi-year partnership to deploy up to 6 gigawatts of AMD Instinct GPUs. Independent analysts estimate the deal at approximately $60 billion over five years, making it the single largest hardware procurement deal among the Magnificent Seven. The deal centers on a custom AMD Instinct MI450 GPU co-engineered with Meta, built on TSMC's 2nm node, featuring 432 GB of HBM4 memory and 19.6 TB/s of bandwidth. The deployment uses AMD's Helios rack-scale architecture, developed jointly with Meta through the Open Compute Project, with 72 MI450 GPUs per rack delivering 1.4 EFLOPS at FP8.

The financial structure is as significant as the hardware. Meta received performance-based warrants for up to 160 million AMD shares (roughly 10% of the company), vesting in stages tied to deployment milestones from 1 GW to 6 GW and AMD stock price thresholds up to $600 per share. This turns Meta from a customer into a strategic partner with a vested interest in AMD's technological roadmap and stock performance.

The first 1 GW deployment begins in the second half of 2026. Each gigawatt is expected to generate double-digit billions in revenue for AMD. If the full 6 GW rolls out by 2030, AMD's Data Center revenue could triple or quadruple from current levels on this single deal alone.

AMD also announced MI350 (shipping now on TSMC 3nm with 288 GB HBM3E) and confirmed MI450 with Helios for second- half 2026. Management claims MI355X delivers 1.6x the memory capacity and 2x the peak FP64 performance of NVIDIA's B200. CEO Lisa Su guided Q1 2026 revenue to approximately $9.8 billion, representing 32% year-over-year growth with gross margins holding near 55%. AMD reports Q1 2026 earnings on May 5.

The Moat Question

The most important variable in this battle is not revenue or margins or even the Meta deal. It is CUDA.

NVIDIA's Compute Unified Device Architecture has been the default programming framework for GPU computing since 2007. Nearly two decades of developer tools, libraries, frameworks, university courses, and enterprise deployments have created an ecosystem that is extraordinarily difficult to replicate. Every major AI framework (PyTorch, TensorFlow, JAX) was built CUDA-first. Every AI researcher trained on NVIDIA hardware. Every enterprise AI deployment was architected around NVIDIA's software stack.

This is NVIDIA's real moat, not the chips themselves, but the software ecosystem that makes switching painful. When a hyperscaler evaluates AMD hardware, the question is not just "is the chip competitive?" It is "can our entire software stack, our tooling, our training pipelines, our inference infrastructure, and our engineers' muscle memory work equally well on ROCm?" The answer has historically been: not quite.

But the Meta deal suggests the answer is changing. Meta did not commit $60 billion to an inferior software ecosystem. The co-engineering of the MI450, the joint development of the Helios architecture through the Open Compute Project, and the multi-generation CPU and GPU roadmap alignment all signal that at least one hyperscaler has decided the ROCm ecosystem is ready for production at scale. If Meta can run its Llama models on AMD silicon at competitive performance, other hyperscalers will take notice.

The CUDA moat is real. But moats erode when customers have a financial incentive to diversify suppliers. At $470 billion in combined hyperscaler AI infrastructure spending in 2026, even the largest buyers have strategic reasons to avoid single-vendor dependence. AMD does not need to match CUDA. It needs to be good enough that the price-performance advantage justifies the switching cost. The Meta deal suggests it has crossed that threshold for at least one of the world's largest AI buyers.

The Valuation Verdict

Valuation is where this battle gets interesting, because the two companies require fundamentally different analytical frameworks.

NVIDIA at $4.82 trillion and a trailing P/E of 40x is priced for perfection, but it is delivering perfection. The forward P/E of 24x on expected earnings growth makes it look reasonable, and the PEG ratio of 0.61 suggests the stock may actually be undervalued relative to its growth rate. The issue is that at $4.82 trillion, the law of large numbers applies. Growing revenue 65% from a $216 billion base means adding $140 billion in new revenue. That is roughly the entire annual revenue of Intel, Qualcomm, and Texas Instruments combined. Can NVIDIA keep growing at this pace? The guidance says yes for at least the next quarter. The question is how many quarters beyond that.

AMD at $588 billion and a trailing P/E of 75-136x (depending on which trailing earnings figure you use) looks expensive on the surface. But the forward P/E of roughly 30x on projected 2026 earnings, combined with 34% revenue growth and the $60 billion Meta deal that has not yet begun contributing revenue, paints a different picture. The market is pricing a company that could plausibly double or triple its Data Center revenue over the next 3-4 years. If the Meta deal executes as planned and AMD captures even 20-25% of the AI accelerator market, the current valuation starts to look justifiable.

The stock performance tells its own story. AMD is up 269% in the past 52 weeks versus NVIDIA's 83%. That 3x outperformance reflects the market's conviction that AMD is on the steeper part of the growth curve. Whether that conviction is correct depends entirely on whether the Meta deal and MI450 deliver what the press releases promise. The first gigawatt ships in H2 2026. That is when the narrative becomes data.

What Could Go Wrong

Risks for NVIDIA

Growth deceleration. Revenue grew 122% in FY2025 and 65% in FY2026. The trajectory is still impressive, but the rate is slowing. If Q1 FY2027 guidance of $78 billion represents a peak or near-peak quarterly run rate, the multiple compression could be severe. At 40x trailing earnings, any hint of a growth ceiling will move the stock.

Customer concentration. Hyperscalers account for just over 50% of Data Center revenue. Those same hyperscalers are actively designing custom AI chips (Google TPUs, Amazon Trainium, Microsoft Maia) and signing deals with AMD. NVIDIA's dominance is not guaranteed indefinitely. Diversification away from NVIDIA is a strategic priority for every major cloud provider.

Export restrictions. U.S. export controls on advanced AI chips to China continue to evolve. China represented a meaningful revenue source before restrictions tightened. Further restrictions could limit NVIDIA's addressable market.

Risks for AMD

Execution risk on the Meta deal. The $60 billion figure is an estimated ceiling, not a guaranteed floor. The warrants vest on delivery milestones and stock price thresholds. If MI450 encounters manufacturing delays, yield issues, or performance shortfalls, the deal could scale down significantly. A 6 GW deployment across four years is an enormously complex logistical undertaking.

ROCm software maturity. The software ecosystem remains AMD's biggest weakness. CUDA's lead is measured in years of developer tooling, not just benchmark scores. If enterprises find that AMD hardware requires meaningful re-engineering of their inference and training pipelines, the performance-per-dollar advantage erodes quickly when labor costs are included.

Dilution from Meta warrants. The 160 million share warrant (roughly 10% of AMD) is a real cost to existing shareholders. If AMD's stock reaches the $600 vesting threshold, those shares represent significant dilution. The deal is structured to align incentives, but the equity component means existing shareholders are paying for this partnership.

Margin gap. AMD's 54% gross margin is respectable, but NVIDIA's 75% gives it enormous pricing flexibility. If NVIDIA decides to compete on price to defend market share, it can cut prices by 20% and still maintain margins higher than AMD's. That pricing power is a structural advantage in any competitive battle.

The Data Picks a Winner

If you are asking which company is the better business today, the answer is NVIDIA and it is not close. Six times the revenue. Twenty-eight times the net income. Twenty-one points of gross margin advantage. A return on equity of 101% versus 7%. The CUDA ecosystem. These are not debatable advantages. They are arithmetic.

If you are asking which stock has more room to appreciate from here, the answer is more nuanced, and it tilts toward AMD.

NVIDIA at $4.82 trillion needs to add the equivalent of an entire Fortune 50 company's revenue every year to justify continued multiple expansion. At 40x trailing earnings, the margin of safety is thin. The stock is not overvalued on a PEG basis, but the base is so large that any deceleration in AI infrastructure spending would be felt immediately. NVIDIA's next earnings report on May 20 is a $78 billion test. If they beat, the stock goes higher. If they miss, the correction could be violent at this valuation. The risk-reward is balanced at best.

AMD at $588 billion is priced for a future that has not arrived but is now contractually committed. The Meta deal is not a hope. It is an SEC-filed agreement with a specific deployment timeline, specific hardware, and specific milestones. The first gigawatt ships in H2 2026. If it works, AMD's Data Center revenue inflects hard upward, the trailing P/E compresses rapidly, and the narrative shifts from "challenger" to "structural alternative." The forward P/E of 30x on 34% growth gives AMD a more favorable PEG ratio than most mega-cap tech companies. And the 269% stock appreciation, while it makes the trailing valuation look stretched, reflects a re-rating that has fundamental backing in a $60 billion deal.

The verdict: NVIDIA is the better business. AMD is the better risk-reward at current prices. If you already own NVIDIA and are looking for a reason to hold, the fundamentals justify it. If you are deploying new capital into the AI chip trade and evaluating where the next dollar goes, AMD's combination of contracted demand, relative valuation, and growth trajectory gives it the edge. The Meta deal is the tiebreaker. It is the first time in the AI era that a hyperscaler has committed at this scale to a non-NVIDIA partner, and the financial structure (with equity alignment at a $600 target) tells you what Meta believes about AMD's trajectory.

At Wealth Engine Pro, we follow the numbers, not the narrative. The narrative says NVIDIA is untouchable. The numbers say NVIDIA is extraordinary but priced accordingly, while AMD is flawed but priced for a future that now has $60 billion in contractual backing. When the gap between narrative and data is this wide, the data usually wins. In this battle, the data gives AMD the edge. Barely.

What Battle Do You Want to See Next?

Battle Stocks is a weekly series. Every week, we put two or three companies in the ring and let the data pick a winner. We want to hear from you: which matchup should we cover next? Cloud infrastructure (GOOGL vs. MSFT vs. AMZN)? Defense primes (LMT vs. RTX vs. GD)? Power generation for AI (NEE vs. CEG vs. Vistra)? EVs (Tesla vs. Rivian)? Enterprise data (Palantir vs. Snowflake)? Tell us on social media or reply to our newsletter, and the best suggestion becomes next week's battle.