As investors debate whether custom AI chips will eventually chip away at Nvidia Corp.’s (NASDAQ:NVDA) dominance, Gabelli Funds portfolio manager John Belton says the latest releases from Meta Platforms Inc. (NASDAQ:META) and Space Exploration Technologies Corp‘s (NASDAQ:SPCX) xAI suggest the chipmaker’s competitive moat remains firmly intact.
In a recent note, Belton pointed to Meta’s Muse Spark 1.1 and xAI’s Grok 4.5 as evidence that frontier AI developers continue to rely on Nvidia’s infrastructure to train their most advanced models.
The AI Race Is Still Running On Nvidia
“Both models were trained on NVDA infrastructure, suggesting there is still a clear value proposition for using NVDA’s stack,” Belton wrote.
The observation comes as investors increasingly question how long Nvidia can maintain its leadership with hyperscalers including Meta, Alphabet Inc. (NASDAQ:GOOGL) (NASDAQ:GOOG), Amazon.com Inc. (NASDAQ:AMZN) and Microsoft Corp. (NASDAQ:MSFT) investing heavily in custom AI chips.
While those in-house silicon efforts continue to expand, Belton argues the latest generation of frontier models shows Nvidia remains the platform of choice for the industry’s most demanding AI workloads.
Read Also: Meta Just Undercut OpenAI and Anthropic by 75%: JPMorgan Says It's More Than a Price Cut
Why Meta Matters
Belton also sees another reason Nvidia has outperformed the broader semiconductor sector to start the third quarter.
He pointed to reports suggesting Meta’s AI infrastructure spending in 2027 could come in well above Wall Street expectations, reinforcing the view that the hyperscaler capital expenditure cycle is far from over.
That matters because hyperscalers account for roughly 50% of Nvidia’s business, according to Belton.
“The market has become concerned about the durability of those revenues,” he wrote, noting that many hyperscalers are operating around break-even free cash flow. Meta’s expanding infrastructure ambitions, however, provide greater near-term visibility into AI spending, even if longer-term questions remain.
Competition May Be Nvidia’s Biggest Advantage
Belton’s most notable takeaway is that Nvidia doesn’t necessarily need one dominant AI winner. Instead, he argues, competition among leading AI labs is a positive.
“Fragmentation in the LLM space is a good thing for NVDA,” Belton wrote, adding that a winner-take-all market for AI models would be less attractive for Nvidia over the long run.
For Nvidia investors, that means every breakthrough from companies like Meta, xAI, Anthropic or others isn’t just another milestone in the AI race—it could also reinforce demand for the infrastructure powering it.
Read Also: SK Hynix vs. Micron: Wall Street's AI Memory Battle Just Got Real
Image via Shutterstock
