Skip to content

Runtime: Jensen Huang puts Nvidia in an impossible situation

+ Revolut's growing pains, MSPs under attack, and Google Cloud's new TPUs.

Runtime: Jensen Huang puts Nvidia in an impossible situation
(Credit: Nvidia)

Welcome to Runtime! Today: Nvidia's perch atop the AI world looks increasingly precarious, Revolut's growing pains, and more.

Please forward this email to a friend or colleague! If it was forwarded to you, sign up here to get Runtime for free every week, or level up here.


First up: Big trouble with Big China

Jensen Huang has always been cut from a different cloth than the rest of the AI industry, and not just because of his trademark leather jacket. Huang grew the chip company he founded over 30 years ago from the shelves of Fry's Electronics into the most important engine of the modern economy while most of the people running the AI model providers were in high school.

But Huang acknowledged Sunday that he had been leading a double life, carrying on an extramarital affair with an executive from a different chip company who appears to have been a spy for the Chinese government. That statement came after Nvidia representatives aggressively defended Huang's interactions with Megaspeed's Alice Huang last week, calling their meetings "not unusual or unexpected" when that obviously wasn't true.

It's not clear what, if anything, Jensen shared about Nvidia or his close friend President Trump during his relationship with Alice, but the fact that we're even having this conversation demonstrates an incredible lack of self awareness from a man who has clearly enjoyed his time in the limelight over the past several years. The New York Times reported Monday that a federal investigation into Megaspeed's possible role in funneling Nvidia chips to China was halted after Nvidia inquired about its progress last year, which will probably prompt further investigation by an incoming Congress that is much less likely to be sympathetic to Jensen's plight than the current one.

And the revelations couldn't have come at a worse time for Nvidia, not that there's really a good time to learn that your billionaire founder and CEO fell into a honeypot. Cloud companies like AWS and Google Cloud have been developing alternatives to Nvidia's ubiquitous AI chips for several years, and the frantic pace of the AI buildout appears to be slowing.

Unless something even more shocking is discovered, Nvidia's board faces a difficult choice about Huang's future given the company's unique leadership structure. But he just added a phrase to the first sentence of his future obituary that will tarnish his legacy, and it might wind up altering the course of enterprise AI.


The rest of The Stack

The price of sand: For a company that so famously preached the strategic importance of elevating software developers, Microsoft is making some curious moves. It will no longer offer developers free sandboxes for building Microosft 365 applications, and unmanaged developer environments are also going away.

Microsoft is nuking free dev sandboxes and hiking costs
A “disgusting sh*t sandwich"? There's always one angry customer…

PSA, MSPs: Companies that sell software to managed service providers are high-profile targets for hackers, given that one successful exploit could ensnare dozens of service providers that have close access to their own customers. That's exactly what happened over the weekend to MSP Process, and the fallout could be huge.

MSP software firm hacked, with customer Entra environments now getting pwned
MFA no help when bad guys have User.ReadWrite.All and authentication method permissions in client tenants

Chipping away: It's looking quite likely that Nvidia is going to be, um, distracted for the next couple of months, which could give Google Cloud a big opening to target enterprise AI customers. On Monday it unveiled two new chips that help relieve a key AI bottleneck and provide extra security to companies in regulated industries.

Google Cloud's new TPUs were designed for context and security
They tackle two enterprise AI concerns: scaling agents eats up memory, and private data needs confidential computing.

Take it to the bank: Like most financial institutions, the Bank of England is treading carefully but steadily into the AI era, according to a recent speech delivered by Chief Data Officer Ben Martin. The bank built a RAG-based workflow that allows employees to send emails when working with central data sources, rather than having to call up a data expert to answer a question.

“Clear financial stability risks” from AI warns Bank of England governor
Central bank itself is using a novel interface to a RAG workflow…

Stack ranking - Enterprise moves

Don Jeter is the new chief marketing officer at Abnormal AI, joining the email security company after almost four years as CMO of Torq.

Mazhar Mohammed is the new vice president of streaming and messaging at AWS, joining the cloud provider after more than 33 years across Lake Washington at Microsoft.


We're also reading:

Total System Failure: U.S. Government Networks Will Need To Be Torched And Rebuilt: Given reports of Russian spyware running on computers in the White House and god knows what else has been going on over the last two years, Karl Bode makes a good argument for the 48th President of the United States to heed.


Thanks for reading — see you Thursday!

Add The Stack on Google