You’d expect Nvidia’s (NVDA) next big catalyst to come from familiar places: quicker GPUs, larger clusters, and more cloud spending.
Google parent Alphabet’s earnings strengthened that assumption.
Google bumped its 2026 CapEx forecast to $195 billion to $205 billion, according to Reuters, underscoring that demand for AI infrastructure remains intense.
Consequently, according to Yahoo Finance, Nvidia (NVDA) closed on Wednesday, July 22, at $212.06, up 2.3%.
Moreover, despite turbulence in tech stocks, Nvidia has delivered a 14.7% gain over the past six months, compared with the S&P 500’s 8.5% gain, according to Seeking Alpha.
However, in a note shared with me, Bank of America argues that Nvidia’s next major opportunity extends beyond the hardware category that has solidified its dominance
Nvidia is pushing into the CPU market, long dominated by Advanced Micro Devices (AMD) and Intel (INTC), where agentic AI could reshape what customers value most.
Chip Somodevilla/Getty Images
Bank of America feels Nvidia’s next big AI catalyst comes from CPUs
Bank of America analysts, led by veteran analyst Vivek Arya, believe the server CPU market for agentic AI is Nvidia’s next major frontier. Moreover, that specific market could expand nearly fourfold to $170 billion by 2030.
More Nvidia:
- Bank of America sees Nvidia’s next $20 billion business
- Morgan Stanley says Nvidia stock remains top pick despite headwind
- Citi sends strong signal to Nvidia investors amid rumors
Nvidia’s Vera CPUs specialize in single-threaded CPU performance, which basically means how quick a processor can complete a specific task at a time.
CEO Jensen Huang directly linked Vera to agentic workloads, saying that,
“AI agents will be the largest users of computing. Vera is the first CPU designed for that future.”
Agentic AI workloads are sequential in nature.
That means each action undertaken by an AI agent depends on the previous one being completed. If the latency in each CPU step can be shortened, the total time taken to complete an agent’s task improves substantially.
Its Vera CPU includes the following:
- 88 custom Olympus ARM-based cores
- 1.2 TB/s of memory bandwidth
- 3.4 TB/s of on-die fabric bandwidth
- A monolithic compute die designed to improve coherence
These features are tailor-made for faster execution, preventing data-access bottlenecks while keeping communications on a single chip to substantially reduce delays.
Nvidia is growing its “AI Factory” moat
Another big point to consider is that Nvidia is looking to control more of the AI data center architecture.
Vera is being co-designed with multiple other Nvidia technologies, which include:
- Rubin GPUs
- Groq LPX processors
- Spectrum networking switches
- BlueField networking and storage products
That’s at the heart of Nvidia’s competitive advantage, which comes from system-level integration, instead of just individual GPU performance.
A customer purchasing an Nvidia stack receives CPUs, GPUs, networking, interconnects, software, and orchestration technologies designed to work together.
That substantially improves performance while bumping switching costs. Moreover, that also helps build the case for Nvidia’s architecture becoming the default platform for developing large-scale AI factories, blowing past its competition in the process.
How does BofA value Nvidia stock?
Bank of America analysts maintain a Buy rating on Nvidia stock, with a $350 price target. At the report price of $207.29, the target implies roughly 68.8% upside.
The price target equates to 26 times 2027 estimated earnings, adjusted for the tech giant’s ballooning net cash position. The multiple sits near the bottom of its historical forward PE range of 25 to 56 times.
Hence, the valuation is near the low end of the company’s historical range, while relying on future earnings growth to drive upside.
Moreover, according to Seeking Alpha, Nvidia stock is trading at 23.6 times forward non-GAAP earnings, 46% below its 5-year average.
With Nvidia’s leadership in two major expanding markets in AI computing and data-center networking, BofA analysts believe it deserves a premium valuation.
Nonetheless, the multiple is moderated by three primary concerns, though:
- AI project lumpiness: Large sovereign, hyperscale, and enterprise projects can be delayed, resized, or recognized unevenly between quarters.
- Gaming cyclicality: Nvidia is exposed to a consumer market that could experience inventory corrections and demand sluggishness.
- Power constraints: Customer demand may exceed the industry’s capacity to build, power, and cool new data centers.
Could AMD’s throughput advantage challenge Nvidia’s CPU push?
According to Bank of America, AMD’s claim could be that Nvidia is optimizing for the wrong constraint.
Nvidia’s focus is on quicker task competition for each individual AI agent, but AMD believes that commercial AI actually resembles an enterprise software platform.
Basically, there are multiple moving parts to that platform, including APIs, databases, middleware, caching layers, and a whole host of other things.
Under that model, it’s important to understand how many agents can a rack potentially support within a fixed budget.
AMD estimates its EPYC 9965 Turin processoris able to deliver nearly 2.4 times the rack-level throughput of Nvidia’s Vera baseline in a modeled 100-kilowatt deployment. Its next-generation Venice platform is expected to take things up a notch or two, increasing that advantage to 3.3 times.
The disagreement also extends to the ARM versus x86 debate.
Nvidia is arguing that the underlying chip standard matters a lot less if its processors and software work in tandem to deliver better overall AI performance. AMD and Intel, however, have one major advantage: most business software was built around their x86 architecture.
For decades, companies have used x86 systems to run databases, security tools, operating systems, and internal applications. That means the software is already tested, familiar, and widely supported.
That might matter more as AI agents work directly with corporate systems, and even though Nvidia offers a tightly integrated AI platform, AMD and Intel stand to benefit from compatibility.
What should investors take away from the CPU battle?
Bank of America doesn’t exactly declare a winner in the Nvidia vs AMD vs Intel debate.
Clearly, we’re still in the early innings of the agentic AI market taking shape, making it too early to conclude which aspects customers will value most.
For Nvidia, the bull case centers on Vera strengthening its full-system strategy, and if consumers prioritize lower latency, tighter CPU-GPU integration, and better utilization of expensive accelerators, Nvidia stands to widen its competitive moat substantially over time.
AMD and Intel, however, have a practical advantage.
Their x86 architecture is deeply embedded across enterprise software, lowering migration costs and making adoption a lot easier for traditional businesses.
For investors, this means the market might not produce a single winner across every workload.
Nvidia may dominate latency-sensitive AI factories, while AMD and Intel remain competitive in enterprise environments built around scale, compatibility, and parallel processing.
Real-world customer deployments over time will provide greater clarity on where the market is heading, helping investors assess how much of that opportunity is already priced into Nvidia’s valuation.
Related: Cathie Wood makes $20.5M bet on sliding tech stock
