Sunday, November 15th 2026
Report: Jensen Huang Eyes NVIDIA Comeback
Jensen Huang has talked with close confidants about returning as NVIDIA's CEO, according to a Sunday report from the Financial Times citing people familiar with the matter. Huang resigned as CEO and left the board on November 12, with CFO Colette Kress stepping in as interim CEO. NVIDIA gave no reason for the exit, which followed reports of an affair with a suspected Chinese spy and a second relationship with an NVIDIA executive who reported to him. Apparently, his Silicon Valley allies have now set up a "war room" to lobby the board to reinstate him and to seek support from the White House.
The "Stand with Jensen" open letter, which recently passed 115,000 signatures when we covered it, now has more than 156,000 names. The campaign says over 7,500 of them are current NVIDIA employees, though signers self-identify and the numbers haven't been independently verified. Elon Musk said on X that he signed it, and Epic Games CEO Tim Sweeney is listed as well. NVIDIA shareholders Brad Gerstner of Altimeter Capital and Gavin Baker of Atreides Management have urged directors not to rush a successor and to keep "all options on the table, including the potential return of Jensen Huang."The decision isn't Huang's to make, though. He owns roughly 3.6% of NVIDIA and never held a controlling stake, so the call rests with the same board that opened an independent investigation into his ties to Alice Huang, a former Megaspeed executive. That probe is reportedly still running. Meanwhile, The Wall Street Journal has reported that NVIDIA, under Huang, funded smaller AI firms so they could in turn buy its GPUs, which adds another wrinkle. NVIDIA reports third-quarter results on November 17, and leadership questions will likely come up on that call.
Sources:
Financial Times, Reuters
The "Stand with Jensen" open letter, which recently passed 115,000 signatures when we covered it, now has more than 156,000 names. The campaign says over 7,500 of them are current NVIDIA employees, though signers self-identify and the numbers haven't been independently verified. Elon Musk said on X that he signed it, and Epic Games CEO Tim Sweeney is listed as well. NVIDIA shareholders Brad Gerstner of Altimeter Capital and Gavin Baker of Atreides Management have urged directors not to rush a successor and to keep "all options on the table, including the potential return of Jensen Huang."The decision isn't Huang's to make, though. He owns roughly 3.6% of NVIDIA and never held a controlling stake, so the call rests with the same board that opened an independent investigation into his ties to Alice Huang, a former Megaspeed executive. That probe is reportedly still running. Meanwhile, The Wall Street Journal has reported that NVIDIA, under Huang, funded smaller AI firms so they could in turn buy its GPUs, which adds another wrinkle. NVIDIA reports third-quarter results on November 17, and leadership questions will likely come up on that call.

248 Comments on Report: Jensen Huang Eyes NVIDIA Comeback
You're absolutely right that the entire system needs to change, but as long as that's not an option, we have to make do with baby steps.
All of what is happening around these subjects was predictable, it was just a matter of time. You can't build an empire on lies and deceit. The truth always wins.
In the meantime, we're just selling ASML machines to you guys :D
What's scary to me in all this is the state of the planet after the dust settles. I don't give a hoot about the idiots who create it, they're already taking far too much attention for their own good.In the trajectory of Nvidia's business the coming years. If there is one thing that's crystal clear, is that Geforce won't be getting a better position than it had under Huang. The company is going to double down on profit before all else, and even with Huang gone, nothing has changed for the better.
Nvidia has pushed gaming ahead immensely under Jensen, and the plateauing of it, is not due to Nvidia, but rather in spite of them. There is simply a limit to graphics. We can shit all over the company for its malicious practices... but that does not erase all that was done for gaming, the two are not mutually exclusive.
On top of that, the quality of the overall Nvidia environment and Geforce products has been stellar over several decades, too.
Fear of the unknown? Of course. This is a high bar to maintain, and exceeding it is hardly even possible. Especially when you consider none of its competitors even remotely came close at any given point in time, and if they did... it was much later and for shorter periods of time until the ball was, again, dropped (it needs no explanation this competitor is AMD and to lesser degree Intel).
Pascal was great, I give you that. If you called that the pinnacle of gaming GPUs, I would agree.
Then came Turing, which is just Pascal +RT +Tensor, so I'll give it a pass.
Then Ampere did absolutely nothing. Nvidia changed the naming convention of a CUDA core, so they could say that you get twice as many while you actually don't. What the actual... Nevermind.
Then came Ada, which brought us... What exactly? Oh I know, high prices even without the crypto bullshit. Awe-fucking-some! Same with Blackwell.
So no, mate. Nvidia has been doing absolutely fuck all that isn't AI related since Turing. Or are you saying that ultra blur version 275.8 and fake frames version 468.4 are such a big technical feat? Come on... I know everybody is pissing their pants about AI now, but who needed this shit originally, seriously? A lame excuse for game developers to forego proper optimisation, that's what it is. Nothing more, nothing less.
So no, I see no revolution here.
Just saying... Pascal wasn't all flowers and sunshine just for gaming... it started the trend
I wouldn't have any problem with the AI crap if it wasn't just an excuse to improve absolutely nothing else while raising prices to high heavens.
Where I worked at the time there was a comical little 'disagreement' between a technical team about who got to use some new machines with pascal GPUs in - some CAD guys wanted them for faster visualisation work, and some project team wanted them to prove/showcase/develop/validate using video inference / ML on recorded video of rail tracks to find damaged fixings.
The CAD stuff would be mildly impacted using older hardware, but the ML/inference work would have been massively slowed down. Who's to say which was more important - one was work for a set of deliverables already bought and paid for by a customer, the other was trying to develop a solution to save money in the future.
We're sort of seeing that argument played out now on a grander scale and, unfortunately, the gamers/CAD group don't have the backing that the 'reduce workplace labour' financing people do.
We can focus on 12VHPWR and/or pricing, or market manipulation / shitty generations of GPUs but that's not the subject and we agree fully.
Technologically, they're ahead of the curve, its just that simple.
It is AI related. At the same time, we have successful upscaling tech because of it that we didn't have prior. There is new lighting tech, that we didn't have prior. I don't recall AMD or Intel pulling that train and I think we can certainly credit Jensen for it, there was a clear push to make some part of the inference tech valuable to graphics processing. Its undeniable that Nvidia was actively looking at making buck on gaming GPUs with, at its core, datacenter technology pushing it forward.
This is what you want to see, from a company, right? Or do you prefer an AMD approach, where the focus is hit/miss, on/off and half of a GPU stack just doesn't get released, or is staggered for so long it misses the upgrade cycle? And the continued focus it DOES have, is enabled in big part by the upscaling tech you just said is not quite relevant in the last years of GPU progress. After all, it is upscaling that enables AMD to keep selling custom APUs that are very cost effective, and THAT is what's keeping their GPU business afloat for the last decade or so. Not Radeon for the PC. You're also casually omitting that a big reason the 9070XT is successful, is because it includes a proper upscaler.
I think you need a reality check here. And then, I haven't even touched on the subject of availability during both the crypto and AI booms. Nvidia delivered, in much better ways than any competitor on that front - simply much more volume and faster time to market.
General shaders, lighting and other concepts predate implementations by vendors.
What made Nvidia finally successful (early days) is the realisation of building hardware to properly align to the software needs (some NV1 failed on spectacularly as they tried to set a standard, and NV3 set the path on) - and ultimately being in the prime position to drive those changes forward pays off... except when you do it wrong (Geforce FX, which actually was a victim of following the implementation too well - AMD's Radeon 9700 just met the minimum DX9 requirement whilst the Geforce FX implemented all FP types but was slower).
Making inference tech valuable to graphics processing is really a return on investment thing - they are ultimately selling GPU dies to people where a large portion of the chip would otherwise have no purpose - putting that to use as a USP is a two-fold benefit in that respect - generating value in having the function there means its not wasted and also can drive sales for those who might find it useful.
For the most part, if they didn't create DLSS FG I doubt their market share would be different bar a percentage point at most.
Its the same thing as Apple vs Microsoft, they both made phones but Microsoft is notoriously bad at hardware and the combo of hardware/software, and so, Apple ran away with the prize and also currently makes the highest quality phone environment, while Windows Phone is in the gutter and Android is the budget and/or homebrew option.
Many deride the company products but the reality is they have actually punched above their weight frequently - hell, being good enough to be able to be in a position of Nvidia - $50, or selling more cores vs Intel and competing that way is good enough going, especially considering how bad things were in the mid-2010s. The others were probably spending more on just one new product development than AMD were spending on a new CPU and GPU release combined. You think they have the same number of software engineers for driver support let alone developer relations?
Why do you think Zen I/O dies and chipsets use IP from others like asmedia for USB / SATA / PCIe connectivity? I'm sure AMD would have loved to do it all in-house for some parts like Intel do for their chipsets.
Really, the fact that the others could have products which could be seen as a mis-execution with the amount of R&D either of their main competitors have is almost staggering in itself.
Great, Nvidia released an improved DLSS version which added extra FG capabilities - what you want me to congratulate them on not f**king up from being in the leading position and having more money than they know what to do with?Not quite - MS just made Windows CE (later Phone) and let device makers do whatever. Until iPhone/Android appeared they were happy to let that train ride as is, but then they realised pulling the software and hardware together would make a better product, and ultimately it was too late - they did introduce some concepts that were copied though so not completely without merit.
Google still kind of actually operate in a middle ground - they make and design for their own hardware but leave flexibility for others to make whatever. Apple never would do that.
I also don't see Android as the 'budget' option... better value sure... I'm not paying a markup for capabilities other already had, but wrapped up in a cosy blanket walking around my walled garden. You can pay for Android devices that are at a near performance parity and probably better hardware.
Likewise, I'm not gonna congratulate Apple on making a bigger fast Mx chip - they have ludicrously more money than Intel or AMD at this point and ONLY design for themselves - frankly if they can't nail it in that position it would be a joke.
I'd say it's very relevant - you can't expect Rolls Royce levels of quality from Dacia, at least not at their first go, and certainly not if they are stuck with engineering their parts from a money pool which is just a fraction in size - it would take forever or until there is no way of improving at the higher end so eventually everyone catches up - much like when we hit the limit of silicon lithography.
Something somewhere has got to give for companies in such a position - I would rather take the fact that the hardware is largely competent as the better approach and wait for the software support.
Nvidia have benefitted from having the money to plough into CUDA as a framework, whereas the earlier AMD Stream framework fell by the wayside as they didn't/couldn't push it in the same way. ROCm is trying to address that same failing but it's not like they can devote equal effort to it.
If all companies had played their cards right in all cases then, yeah for sure, we would have a different marketplace which would probably either have more direct competition and parity - but we are in a capitalist society so odds are equally good just one mega conglomerate dominating.
They messed up moving to new processes on the fab side and just kept low effort evolving the CPU side whilst living off of 14nm refinements and minor arch improvements for over half a decade.
14nm++++ is more than just a meme, and points to how stalled things were at Intel whilst AMD were getting 7nm parts from TSMC.
Zen was a massive improvement on the AMD side jumping the equivalent of 5+ years of stalled primary architecture improvements instead of living off of incremental changes to bulldozer. Even Bobcat to Jaguar was a much better improvement than what was going on with the dozer lineup.
But Zen was still behind Intel, who were mid being stuck at 14nm. This allowed AMD to iterate ahead.
We're about to see if Intel & AMD can maintain a performance level, seeing as they are more or less getting an even performance point in terms of IPC and just separated by execution and timing.
As for ARM, they are mainly an IP shop and have made a good business of doing it and essentially their revenue funds their development with less costs on production, etc.
And they have taken over the world - there are more ARM powered CPUs than there are x86. Big vs small CPUs is a matter of people being willing to manufacture them (Intel made small Atom / AMD made small Bobcat/Jaguar, whereas ARM companies like Ampera or Google make big chips for server) - it wasn't until data centres wanted to go down more efficient high scaling systems that pushed ARM much harder into that sphere, combined with better software stacks and development leveraging many cores, but suitability for workloads and developer familiarity are key. Don't forget ARM started on the desktop but their architecture was leveraged more towards low power since then - it wasn't until the last decade or so that 'big' ARM CPUs have started to be a more common thing.
They (or really their customers / instruction developers) also really benefitted from AMD being not competitive (on a perf/watt level) and Intel stagnating, but from an overall platform offering standpoint they are not as well positioned.
Ps. I don't know why I get the feeling of desperate excuses from Nvidia fans who have had the rug pulled out from under their feet and are in free fall.
I'm not even going to open a discussion on this, as nothing will convince me otherwise. What saddens me even more is that AMD is following this bullshit instead of making something unique and better.I don't care about the product stack. I only care about the single product I'm about to buy. The 9070 XT is a good card, and at launch, it only cost me just a hair over £500. That's what matters. I don't give a rat's arse that there's no 9099 XTXX Ubermensch edition, as I wouldn't buy one anyway. Super expensive luxury toys don't interest me. Raising the performance bar while also raising prices isn't progress. Raising the performance bar only in the high end isn't progress, either.
I'm sick of people applauding Nvidia for all the shady crap they've been pulling in recent years, and calling it technological progress while all Nvidia ever wanted is to push AI down our throats with proprietary technologies that no one ever wanted, but suddenly, everybody is hooked on for some reason. While I have my doubts, I hope this will change direction, if ever so slightly, without Jensen. The guy is fake as fuck, and all he cared about in the last decade is AI. I'm glad he's gone. That's that.
This is not because they loved to stop progressing.
This is ALL about design wins, and that is EXACTLY what Nvidia has achieved under the CUDA moniker, and now with RTX. Upscaling on a part of the GPU hardware, is a design win, its as simple as that, and the rest of the industry followed suit - but their implementation was lacking and this has been a major influence on what to buy.
We can talk for ages about who's to blame, but that game never stops and none of these companies would have won anything if it weren't for actual, tangible design wins. We buy products because they are well designed, and competitive - not because company B didn't get the same amount of money for R&D or got wiped into a corner on the sales market. Those factors are never the cause, they're effects of a cutthroat marketplace. Nowhere has this stopped companies from having a great idea and making it happen. Great ideas MAKE companies.Interesting, the market says you are dead wrong. You even confirmed it yourself by buying the AMD product with the best version of its upscaler recently. AMD, itself, was and still is pretty big on it, too - it even drives their console business, which in turn feeds all our PC gaming business.
GL with that rationale of yours.