By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel agrees that Nvidia's CPU ambitions, driven by agentic AI, present a significant growth opportunity, but execution risks and potential margin compression are substantial concerns.
Risk: Execution risk and potential margin compression due to CPU mix shift
Opportunity: Expansion of total silicon TAM beyond $1T through additive CPU revenue atop sustained GPU demand
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Investors are gearing up for Nvidia's (NASDAQ: NVDA) fiscal 2027 second-quarter earnings, which will be released after the market closes on Aug. 26.
Nvidia is at the center of the artificial intelligence (AI) trade, which is itself at the center of the broader market right now. So all investors will be at least somewhat interested in how the company performs and what CEO Jensen Huang has to say about overall demand.
Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »
Nvidia-specific investors will likely focus closely on what Huang has to say about his previous projection that the company will generate $1 trillion in sales from its Blackwell and Vera Rubin graphics processing units (GPUs) between 2025 and calendar year 2027.
While it's a massive number, there is another emerging part of the business that investors shouldn't overlook.
While Nvidia doesn't break out its revenue by specific chip model, all GPU revenue falls within the company's data center division, which now makes up the bulk of Nvidia's revenue.
Nvidia's fiscal year begins in late January or sometimes early February, so it is fairly close to the calendar year. In fiscal 2026, most of which occurs in calendar year 2025, Nvidia's data center division generated nearly $194 billion in revenue.
According to Visible Alpha, Wall Street consensus estimates expect data center revenue to be nearly $368 billion in fiscal 2027 and nearly $531 billion in fiscal 2028.
Adding up all three years comes to roughly $1.09 trillion, so heading into the upcoming report, investors will be razor-focused on whether fiscal 2027 and fiscal 2028 data center revenue projections increase or decrease.
While the focus will and should be on the GPU business, investors should not overlook Nvidia's central processing unit (CPU) business, which is fairly new. CPUs, which power older consumer electronics such as phones and laptops, have seen a major resurgence due to agentic artificial intelligence.
AI agents are used to perform autonomous tasks with minimal human interaction. While GPUs continue to power the reasoning, CPUs are now viewed as critical for actually performing the task, whether it's planning how to carry it out or accessing external files or data sources.
Last quarter, Nvidia launched its Vera CPU, specifically built for agentic AI. On the company'searnings call Huang said this opens up a $200 billion total addressable market (TAM), and Nvidia expects about $20 billion in CPU revenue this year alone, instantly making it one of the largest, if not the largest, CPU players.
"With respect to CPU use, an agent is essentially what people call a harness," Huang told analysts on the company's most recentearnings call "And so the harness runs on CPU... The world has a billion human users. I sense that the world is gonna have billions of agents. Not today. I mean, we are gonna grow into it."
It's not just Huang who believes the CPU opportunity is massive.
Intel's CEO, Lip-Bu Tan, has also publicly said that the ratio of CPUs to GPUs used for AI inference has already moved from 1:8 to 1:4 and could approach parity or better in the future.
In a research note from June, Bank of America analysts expect the CPU TAM to grow fivefold from $35 billion in 2025 to $170 billion by 2030.
Pegging down a GPU TAM is more difficult, but Bloomberg projected earlier this year that it could rise to $486 billion by 2033, assuming a 14% compound annual growth rate (CAGR). That means by 2030, it would have been $328 billion, still 92% higher than the CPU TAM. But who knows how things will change, especially if Tan is right?
Either way, investors should look for any updated guidance and comments from Huang on the CPU opportunity, as it looks to be an important driver for the company and AI trade as a whole moving forward.
Before you buy stock in Nvidia, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Nvidia wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $411,427! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,335,252!
Now, it’s worth noting Stock Advisor’s total average return is 965% — a market-crushing outperformance compared to 215% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of August 12, 2026. *
Bram Berkowitz has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Nvidia. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"Nvidia's CPU foray is a legitimate but still unproven $20B+ incremental revenue stream that investors should monitor rather than celebrate prematurely."
The article correctly flags Nvidia's (NVDA) emerging CPU business as an under-appreciated growth vector, with Huang's $20B revenue guide this year and a new $200B TAM for agentic AI. Data-center revenue is already projected to exceed $1T cumulatively through FY28, so incremental CPU upside is real. However, the piece glosses over execution risk: Vera Rubin CPU is brand-new, faces entrenched x86 competition from Intel/AMD, and the claimed 1:4 to 1:1 CPU:GPU inference ratio remains speculative. NVDA's forward P/E of ~38x on 2026 EPS already prices in heroic CPU share gains that may not materialize at scale before 2027.
If agentic AI adoption disappoints or Intel successfully defends its CPU moat, the $200B TAM could prove aspirational and Nvidia's CPU revenue might undershoot $20B, forcing multiple compression on the core GPU story that still drives 90%+ of profits.
"Nvidia's expansion into CPUs is a strategic defensive move to control the entire AI compute stack, though its impact on long-term gross margins remains the primary risk factor."
The article's pivot toward Nvidia's CPU ambitions is a critical, albeit under-discussed, hedge against GPU commoditization. By positioning the Vera CPU as the 'harness' for agentic AI, Nvidia is attempting to capture a higher share of the inference stack, effectively verticalizing the AI infrastructure. However, the $200 billion TAM figure Huang cites is speculative; it assumes a rapid, frictionless transition to agentic workflows that may face significant latency and integration bottlenecks. Investors should watch for margin compression as Nvidia shifts from high-margin H100/Blackwell GPUs to a more hardware-diverse portfolio, which could pressure the current 70%+ gross margin profile if CPU manufacturing costs scale faster than software-driven synergies.
Nvidia's push into CPUs risks alienating key partners like ARM and could trigger a defensive consolidation from Intel and AMD, who are already optimizing their own x86 and custom silicon for agentic workloads.
"Nvidia's CPU business is real and material, but it's a margin-dilutive growth vector that the market hasn't fully priced as a headwind to the GPU-centric bull case."
The article conflates two separate narratives: GPU TAM saturation risk and CPU opportunity emergence. Nvidia's $1T GPU projection (FY27-28) is already baked into consensus; the real question is whether execution matches it. The CPU angle is genuinely novel—$20B revenue this year on a $200B TAM is credible given agentic AI's infrastructure needs. But the article glosses over a critical tension: if CPUs-to-GPUs ratios shift from 1:8 toward 1:4 or parity as Intel's Tan claims, that's margin compression for Nvidia's higher-margin GPU business, not pure upside. The $531B FY28 data center estimate assumes GPU dominance persists. Any material CPU mix shift could underperform that consensus without being 'bad news.'
If agentic AI adoption accelerates CPU demand faster than GPU demand, Nvidia's blended gross margin (currently ~75% on data center) compresses significantly, and the $1T projection becomes a revenue mirage masking margin deterioration. Wall Street's FY27-28 estimates may already price in modest CPU contribution but not a structural shift in the GPU/CPU ratio.
"Nvidia's Vera Rubin CPU opportunity could unlock a sizable, durable revenue stream that diversifies the business and supports a multi-year earnings re-rating as AI adoption broadens."
The headline update on Nvidia’s $1 trillion GPU opportunity remains central, but the article emphasizes a second growth leg from Vera Rubin CPUs and agentic AI. In the near term, the CPU TAM and Vera Rubin revenue are plausible catalysts if adoption accelerates, with data-center demand likely to be a multi-year driver and potential margin upside from software, services, and ecosystem lock-in. However, execution risk is non-trivial: Vera Rubin is still nascent, CPU margins may lag GPU, the AI agent stack requires broad software adoption, and competition/regs could temper upside. If CPU adoption stalls or supply constraints bite, the implied re-rating could unwind despite GPU strength.
The CPU opportunity is highly aspirational: Vera Rubin is early-stage, CPU margins are uncertain, and the AI agent software ecosystem may struggle to gain broad traction, risking a disappointing revenue pull compared to the GPU-led upside.
"Agentic workloads are likely additive to GPU demand rather than substitutive, expanding overall TAM."
Claude's margin-compression warning assumes CPU mix directly displaces GPU units, but the 1:4 inference ratio Huang cited implies additive CPU revenue atop sustained GPU demand for training and high-end inference. Nobody has flagged that agentic AI could expand total silicon TAM beyond $1T, letting Nvidia harvest both without the either/or trade-off the panel assumes.
"Finite data center power budgets mean CPU growth will structurally cannibalize, rather than add to, high-margin GPU deployment capacity."
Grok, your 'additive' theory ignores the physical reality of data center power constraints. Power budgets are finite; every watt diverted to a CPU is a watt unavailable for a GPU. If agentic AI requires a 1:4 CPU-to-GPU ratio, that is a structural cannibalization of the most profitable silicon in the rack. We aren't just looking at margin mix shift; we are looking at a hard ceiling on the total compute density Nvidia can deploy per rack.
"Power constraints matter only if data center budgets stay flat; Nvidia's thesis requires new capex, not reallocation."
Gemini's power-budget constraint is real, but Grok's additive case isn't dead. Modern data centers are already power-limited; Nvidia's pitch assumes customers *upgrade* total rack power for agentic workloads, not reallocate existing capacity. The question isn't cannibalization within a fixed power envelope—it's whether enterprises will capex more to run both training GPUs and inference CPUs. That's a demand question, not a physics one. If they won't, Gemini wins.
"Power is not a fixed ceiling; TAM depends more on demand timing and software adoption than on a hard watt constraint."
Gemini, your power-budget concern paints a hard ceiling on rack density, but that assumes fixed capex. In reality, customers often expand total rack power and cooling when AI demand rises, so CPU-GPU mixes could rise without a strict watt constraint. The bigger risk remains demand timing, software ecosystems, and supply constraints; power is a factor, but not a guaranteed limiter to Nvidia's TAM or margin trajectory.
The panel agrees that Nvidia's CPU ambitions, driven by agentic AI, present a significant growth opportunity, but execution risks and potential margin compression are substantial concerns.
Expansion of total silicon TAM beyond $1T through additive CPU revenue atop sustained GPU demand
Execution risk and potential margin compression due to CPU mix shift