My years at Goldman Sachs and KPMG taught me something the nerd in me rarely understood: decisions at the top usually need more than a reason. They need an incentive — someone powerful’s, or enough people’s together. This incentive-driven market theory of mine tells me that political machines do not move unless there is an incentive to move.
They sure moved fast against Jensen Huang, former CEO of NVIDIA.
The allegation involving a suspected spy was a minor box-office success at best. The employee story gave them better justification. But like most real people around me, I have zero qualms about consensual relationships. A man over 60 and a suspected spy may grab attention. But no one is boycotting NVIDIA. The orders are not cancelled. It is business as usual. As far as numbers are concerned: nothing happened.
Why did the board that initially backed him withdraw its support? The second story? But was it a secret or an open secret? Again, is NVIDIA losing contract because of it? Then what might have caused the flip?
My mind goes to Washington. Could talks of “treason” or a “national security threat” have sealed the deal? Could someone, or some group, have wanted Jensen out? But why? Who benefits?
Keep reading to see if you agree.
Now some reasonable readers may already be saying, “This isn’t my first rodeo. Jensen will be back in a couple years. Maybe less.” Nope, I would not bet on it.
A corporate scandal? Yes, I would agree with those readers. But it is a whole new game when politics gets involved. When the machines move, they have a target to secure. So what is the target?
We can only speculate. But a federal board or agency overseeing NVIDIA’s exports sounds a very reasonable next step. Whoever controls that oversight would have enormous influence over the global chip supply. An invisible emperor, or “Silicon Lord.” A perfect retirement for a career politician.
As the CEO and founder of a startup, I think about what this means for my own business. I am a U.S. citizen building a company in the U.S., making a custom tensor engine and robotics models. So I hope my domestic chip supply stays safe, without too much inflation.
But what about other countries and universities around the world? Will acquiring NVIDIA chips become harder, easier, or stay unchanged?
My first bet is Apple. I think more universities and small-to-medium companies will turn to Apple Silicon. These are consumer grade machines. Securing one does not take a special license. And they are becoming good enough to compete not just on performance per watt, but on raw performance.
That trend was already underway. I think recent events will only accelerate it. Robotics models running directly from a MacBook to physical robots, completely bypassing NVIDIA boxes. M5 Ultra machines powering commercial servers for inference and training. Macs are getting genuinely competitive for serious model workloads. I am not talking about hobbyist home hosting. I mean real commercial services.
Google will also benefit. Google has not been quiet. It has been busy rolling out global-scale fiber networks powered by its Palomar Optical Circuit Switches and Virgo network fabric. Its distributed AI network is capable of training models across data centers. Google is ready to step into whatever vacuum NVIDIA leaves behind.
Broadcom must be happy about this development as well, along with inference chipmakers like Groq and Cerebras.
But who is really smiling? AWS.
Its custom Trainium/Inferentia roadmap has had rough patches, and its next-generation rollouts have not always stolen the spotlight. But its existing fleet works at scale. Amazon wants cloud computing to own everything AI related.
However, Microsoft may turn out to be the biggest winner of them all. Those who cannot secure NVIDIA chips directly will turn to Microsoft Azure, where they can rent capacity instead. Blackwell today, Rubin as it rolls out. My bet is that Microsoft’s allocations will put it in a position to serve major global players.
Now here is where my incentive-driven theory takes me. Washington does not see China slowing down on AI. The chip deal had failed. The Chinese took it as the perfect storm to nurture their domestic chip manufacturing.
What cards does Washington have over Beijing now? To sell NVIDIA chips at a premium? To restore dependence on U.S. chips? My theory predicts the think tanks want China and the rest of the world to develop and run their AI on U.S. controlled clouds rather than own standalone NVIDIA chips.
That would be the moat. Securing it could be the “official target” behind Operation Jensen Out, regardless of the personal incentives of those who make the decisions.
I personally like Jensen. He is one of the leaders I respect. I admire his practical approach, straightforward personality, and his grit. The headlines read like Austin Powers. We gossip over it, but none of us are outraged. His private life is his own after all. My concern is what comes next.
What will happen to the global chip allotment?
Who will sit on the empty throne?
Press enter or click to view image in full size