Silicon Valley CEO Says Nvidia Has One Weakness — and He Plans to Exploit It

The artificial intelligence boom has created one of the most dominant technology companies Wall Street has ever seen. Nvidia (NASDAQ:NVDA | NVDA Price Prediction) has become the backbone of modern AI infrastructure, growing from a gaming-chip specialist into a company worth trillions of dollars. Its data center revenue exploded from $15 billion in fiscal 2023 to almost $194 billion in fiscal 2026. 

Yet history suggests no technology leader stays on top forever. IBM (NYSE:IBM) lost ground to personal computers. Intel missed the mobile revolution. Cisco (NASDAQ:CSCO) never fully capitalized on cloud computing. Now one Silicon Valley CEO argues Nvidia’s greatest strength may also be its biggest weakness.

The Intel Comparison Is Hard to Ignore

Appearing on the All-In Podcast, Cerebras Systems (NASDAQ:CBRS) CEO Frank Bruno argued that incumbents often struggle when computing architectures change. His example was Intel (NASDAQ:INTC), which dominated traditional PCs but failed to adapt when smartphones shifted computing toward lower-power mobile processors.

Bruno believes AI is approaching a similar transition.

Nvidia’s current architecture was built to excel at AI training — the computationally intensive process of teaching large language models how to perform tasks. That market created enormous demand for the company’s GPUs and software ecosystem.

According to Nvidia’s fiscal 2026 results, data center revenue represented 90% of total revenue, highlighting how closely the company’s fortunes are tied to AI infrastructure spending.

Bruno’s argument is that the next phase of AI will increasingly focus on inference — the process of running trained models in real time. If inference becomes the larger market, architectures optimized specifically for that workload could gain an advantage.

Let’s be clear: that does not mean Nvidia is destined to lose. Intel’s mistake was not simply being large. It was becoming optimized for yesterday’s computing paradigm.

Cerebras Is Betting on a Different Architecture

Cerebras has built its business around a radically different approach. Rather than connecting thousands of smaller chips together, the company developed wafer-scale processors that use an entire silicon wafer as a single chip. According to Cerebras product documentation, its latest Wafer Scale Engine contains trillions of transistors and hundreds of thousands of AI cores on a single device.

The goal is simple: eliminate communication bottlenecks that occur when AI workloads are spread across large clusters of chips.

Here’s how the two approaches compare:

Investors initially embraced that vision. Cerebras completed its IPO earlier this year, and the stock jumped sharply during its first trading session — closing 68% above its offer price — before giving back a portion of those gains in subsequent weeks. It currently trades just 22% above the $185 offer price.

That volatility reflects a familiar reality: disrupting an industry leader is much harder than identifying a theoretical weakness.

Nvidia’s Response May Validate the Threat

Surprisingly, one of the strongest arguments in Cerebras’ favor may be Nvidia’s own behavior.

The company is pursuing technologies designed to improve inference performance and reduce the communication overhead that emerges when AI models run across multiple processors. Nvidia’s efforts to incorporate architecture concepts similar to those championed by inference-focused competitors suggest management recognizes the opportunity.

That does not mean Cerebras will win. Nvidia generated nearly $194 billion in annual data center revenue last year — it was up 92% in FY2027 Q1 — and enjoys one of the strongest software ecosystems in technology through CUDA. Those advantages create powerful barriers to entry.

Granted, technology transitions can happen faster than investors expect. Intel learned that lesson during the smartphone era. Whether AI inference becomes a similar turning point remains one of the most important questions in the industry.

Key Takeaway

In short, Frank Bruno’s thesis is not that Nvidia is weak today. The numbers show the opposite. His argument is that every dominant platform eventually becomes optimized for the world that created it.

Cerebras is betting that AI inference will become the next architectural battleground and that its wafer-scale design is better suited for that future. Nvidia’s scale, software ecosystem, and revenue base make it the clear leader today. Yet smart investors should pay attention whenever a challenger identifies a specific technological shift rather than simply claiming it can compete on price.

Ultimately, the most interesting part of Bruno’s argument is not that Nvidia has a weakness. It is that he has identified exactly where he believes that weakness will emerge.