Listen to the audio version of this article (generated by AI).
In July 1961, John Profumo, Britain’s secretary of state for war, met a 19-year-old model at an English country estate.
While Profumo was a powerful man with access to sensitive military information, the model, Christine Keeler, knew another man with an interest in those secrets.
His name was Yevgeny Ivanov, a Soviet naval attaché and intelligence officer.
When Profumo and Keeler’s brief affair became public, the connection to Ivanov turned a private affair into a Cold War security scandal.
By the end, Harold Macmillan’s government and the public trust were badly shaken.
More than six decades later, another affair has raised questions about a powerful man’s exposure to foreign intelligence… this time at the center of the AI economy.
Nvidia (NVDA) CEO Jensen Huang sent his employees an apology Sunday acknowledging being unfaithful to his wife.
“I was unfaithful to my wife,” he wrote, acknowledging the damage to his family and pledging to cooperate with an independent investigation into national security concerns raised in news reports.
Nvidia said its board had no reason to believe company secrets or national security had been compromised. But for investors accustomed to watching the leather jacket take the stage with another AI breakthrough, it was a jarring development.
But the questions now extend beyond Huang’s personal conduct to Nvidia’s leadership, its China business, and its relationship with Washington.
That’s why I’m not putting any more capital to work in Nvidia right now.
Just last week, Nvidia was one of the stalwarts to lean into. Yes, its business remains strong, but the uncertainty surrounding that business has increased, and we want answers before recommending investors buy more.
Some commentators see consequences reaching far beyond Huang. Such as The American Prospect, which published an article titled “The Honeypot That Might End the AI Boom,” arguing that Nvidia’s central role in supplying chips and financing AI companies could send shockwaves throughout the industry.
I understand that concern. If Nvidia suffered serious operational or financial damage, the fallout could spread.
But I think investors should consider another possibility… That this controversy could ultimately help broaden the AI trade.
That may sound counterintuitive. Nvidia has become so closely associated with the AI boom that a problem for its CEO can feel like a problem for the entire industry.
But ask yourself…
Does this controversy make companies less interested in AI coding tools? Does it eliminate demand for AI agents? Does it reduce the computing power needed to train and run advanced models?
I don’t see why it would.
What it could change is where customers choose to spend their money.
And if demand stays strong while Nvidia’s grip on that spending loosens, competing chipmakers could gain an opening.
To be clear, that possibility isn’t an all-clear signal to buy AI stocks. Financing pressures, supply shortages, and the capex slowdown still matter. We’re waiting for the fundamentals and the charts to justify getting more aggressive.
But we can be cautious about Nvidia today while recognizing that a less concentrated AI market could create more winners over time.
Here’s how that shift could unfold…
The Spending Could Spread Beyond Nvidia
My base case is that the financial impact on Nvidia remains manageable and Huang continues leading the company. That said, I have no idea what the investigation could uncover.
Under that scenario, the controversy could give major customers another reason to accelerate efforts to diversify their chip supply.
Perhaps Microsoft (MSFT) moves faster on its own chips.
Perhaps Meta (META) directs more business toward Advanced Micro Devices (AMD).
Perhaps Amazon (AMZN) expands Trainium deployments, or Alphabet (GOOG, GOOGL) leans further into its own processors.
These are, of course, theoretical outcomes. But they illustrate the opportunity that AI spending can shift between suppliers without disappearing.
That shift could also reach beyond the main processor.
For instance, Nvidia supplies a broad collection of technologies that work together. When customers choose different chip platforms, they may create additional opportunities for outside networking, connectivity, and optical suppliers.
So let’s explore three chip companies that could benefit from this broader distribution of demand.
Advanced Micro Devices: An Alternative for AI Buyers
First, we have Advanced Micro Devices (AMD), which offers the most straightforward version of this thesis.
Its Instinct accelerators compete for the computing work required to train and run AI models, supported by AMD’s ROCm software platform. AMD also offers networking and server technologies for large AI systems. If a major customer wants to direct more of its AI budget toward another supplier, AMD gives it an alternative.
That is why I specifically raised the possibility of Meta going further toward AMD. The opportunity would come from winning a larger portion of spending that customers already intend to undertake.
AMD wouldn’t need the entire industry to abandon Nvidia. Rather, a customer allocating more of its next deployment to AMD could be meaningful.
Of course, reputational concerns alone won’t determine a multibillion-dollar purchasing decision. AMD still has to deliver the performance, software support, reliability, and economics customers need.
The controversy could increase customers’ willingness to consider alternatives. AMD would then have to convert that interest into orders.
Broadcom: Helping Customers Build Their Own Chips
Broadcom (AVGO) offers a different route to the same outcome.
Some large AI customers want more control over the chips powering their operations. They can pursue processors designed around their particular workloads, with help from a semiconductor partner.
Broadcom already participates in that shift. Its expanded partnership with Meta covers custom AI chips and the networking technology supporting their deployment.
That helps explain why a cloud company’s decision to build more of its own chips can still create an opportunity for public-market investors. “Our own chip” does not necessarily mean every part is designed and delivered internally.
Specialist partners can play substantial roles.
My thesis is that added concern about dependence on Nvidia could strengthen the incentive to invest in those alternatives. Broadcom could benefit if customers accelerate existing programs or commit more spending to future generations of custom chips.
These are lengthy development decisions, so I wouldn’t expect a weekend controversy to produce an immediate wave of revenue.
But over the coming quarters and years, faster diversification could matter.
Marvell Technology: Supplying the Chips and Connections
Marvell Technology (MRVL) is another company I mentioned because its opportunity extends across custom processors and the connections that allow AI systems to function.
Marvell helps cloud and data center customers develop custom chips. Its capabilities also include the technology that moves data between processors, memory, and other parts of an AI system.
That second piece deserves attention.
An AI data center needs thousands of components to work together. Choosing a different processor can create opportunities for other suppliers throughout the surrounding system.
This is the broader effect I was describing on the podcast. If more demand flows toward alternative platforms, companies supplying the supporting technology could participate, too.
For Marvell, that could mean opportunities in both custom computing and connectivity.
The important word is could. Customers still need to select its technology, complete development, and move into production. Where diversification creates a potential opening, execution determines how much business Marvell captures.
Stay Patient and Look Beyond Nvidia
There is an important limit to this argument.
If the controversy seriously damages Nvidia’s operations or its ability to finance AI projects, the consequences could spread throughout the industry. And even if the fallout stays contained, financing costs, Washington, and supply shortages still threaten to slow the buildout.
That’s why we’re staying patient. Forget about buying Nvidia for now — wait for spending expectations and the charts to give us a stronger foundation before getting aggressive.
But patience on AI infrastructure doesn’t mean putting all our research on hold.
The opportunity we’ve discussed today comes from following where spending could go next. AMD, Broadcom, and Marvell could benefit as customers diversify their chip purchases. Elsewhere in technology, a different shift could create opportunities in the companies that help money move.
That’s the focus of my research into Elon Musk’s ambitions for payments and financial services.
His vision reaches into something people do every day: sending money, making purchases, and managing their finances. Bringing those activities together would require technology and financial infrastructure… and could create business for the companies helping make it possible.
For you, the question is which publicly traded businesses could benefit if that vision gains traction.
I’ve been studying those potential beneficiaries and the roles they could play. It’s a separate investment thesis from the AI data center buildout, and one worth understanding while we wait for that trade to stabilize.
