Intel Weighed $20 Billion Nvidia Takeover in 2005 (nytimes.com)
- Reference: 0175315725
- News link: https://slashdot.org/story/24/10/24/1143253/intel-weighed-20-billion-nvidia-takeover-in-2005
- Source link: https://www.nytimes.com/2024/10/24/technology/intel-ai-chips-mistakes.html
[1] https://www.nytimes.com/2024/10/24/technology/intel-ai-chips-mistakes.html
[2] https://slashdot.org/story/24/08/01/2126209/intel-to-cut-16000-jobs-to-save-costs
And if I had just... (Score:5, Insightful)
Bought pre-iPhone Apple stock or mined a bunch of Bitcoin in the early 2000's....
Re: (Score:2)
Rockstar games was almost bought by a different game publisher. I forget which one but it was a choice between them and something like infograms and you can tell how that turned out
Re:And if I had just... (Score:5, Interesting)
> Bought pre-iPhone Apple stock or mined a bunch of Bitcoin in the early 2000's....
Umm... bitcoin didn't exist in the early 2000's. I believe it came into existence around 2008/2009.
I actually mined Bitcoin, I think I had something like 1.2 BTC back in 2011, then I lost the wallet. It was worth next to nothing at the time. Now it's enough to buy a nice car.
But the better analogy would have been to talk about BlockBuster not buying Netflix back in the day...
NVIDIA went to AI because of AMD (Score:5, Informative)
NVIDIA has a history of doing something different every time there is competition. When it was raster performance in games, NVIDIA was in the lead, then AMD showed signs of being competitive, so NVIDIA made ray tracing the focus, then, AMD started to show some competence with ray tracing, so NVIDIA switched direction to AI. If MI350 or 400 turns out to be competitive when it comes to AI performance, NVIDIA will switch directions again.
The Intel board was correct, Intel buys companies like Altera, but then doesn't know what to do with the purchase.
Intel is a real "passion fingers" company (Score:2)
Because it fucks everything it touches. The intel board was probably right.
An interesting counterfactual (Score:5, Insightful)
It's interesting to speculate about, but too far back in the timeline to say how things would be today.
It certainly would *not* be the case that everything else would played out the same - NVidia's increasing dominance in GPUs, then becoming the de-facto for machine learning (CUDA's first stable release wasn't until 2007), then the bedrock of current AI stuff, leading to a $3T market cap - but just with Intel as the owner. Mergers and Acquisitions simply don't work that way.
Poison pill (Score:2)
The rumor at the time was that the requirement is that the CEO of NV becomes the CEO of the merged companies, and Intel wasn't going to do that (who would!).
Lazy Board (Score:2)
> The board rejected the proposal, citing Intel's poor track record with acquisitions
Others have already commented on how nVidia would not be where it is today as part of Intel, but this Board admission is somewhat stunning.
They were dissing their management's skill without doing something about it.
It was only a skunkworks at Intel Israel that dug them out of their P4 Hell with an efficient die-shrink PIII (Core/Core2).
A situation like that is where the Board has a legal responsibility to do something.
No
Smarter move (Score:2)
Instead of acquiring it, the board members (probably) invested in NVIDIA as individuals.
Tables turned (Score:2)
How long until nVidia weighs taking over Intel?
At least their board is insightful (Score:5, Insightful)
If Intel had bought Nvidia they would have tanked both companies.
Re: (Score:2)
Not necessarily. The fact is, though, in 2005, nobody was doing AI with NVidia chips.
Re: (Score:2)
Sure, this would have been about having an even tighter duopoly over core computer components.
"Intel integrated graphics" wasn't a gamer upsell, and Intel knew it.
Re: (Score:2)
> Not necessarily. The fact is, though, in 2005, nobody was doing AI with NVidia chips.
Doesn’t matter. The CEO at the time, saw something there. Something investible.
Now I hope that CEO is retired at home reading this mornings Slashdot post, and having one hell of a laugh. It’s a shame he passed before that happened.
Re: (Score:2)
Comptuer Vision was being done on the GPU back then. The OpenCV library has been around for 20+ years, and it was always the dream to offload some of it from CPU to other hardware like a DSP, GPGPU, or FPGA.
But yeah, CV with Deep Learning to identify the kind of object instead of simply the boundry of an object, didn't show up until around 2012 or so. Some of them accelerated with GPUs shortly after their introduction.
Someone at Intel would have had to predict the path for technology 5-10 years into the fut
Re: At least their board is insightful (Score:2)
Not a whole lot of things beside graphics were done on GPUs in 2005. This was 2 years before CUDA released. So that's still the cg era when getting any kind of gpgpu computation was an amazing feat.
Though I suppose you could possibly do convolutions and color histogram with cleverly designed shaders.
Re: (Score:2)
Jensen tanked Nvidia twice, had to be bailed out by big pocketed investors. Jensen believed in quadrilaterals as primitives, whereas the industry was going forward with triangles. A former boss from LSI logic, Tom Perkins, gave Nvidia a $20m lifeline. Which they would go on to run through and come up with a design for Sega that they would eventually not use, yet pay $5m for anyways (basically a donation to Nvidia).