Bill Gates says Intel has lost its way, fallen behind in chip design and fabrication

midian182

Posts: 10,445   +139
Staff member
Big quote: The last few years have not been kind to Intel. The company has seen its fortunes fall as rivals continue to make great strides, both financially and technologically. In a recent interview, Microsoft co-founder Bill Gates shared his thoughts on the situation, stating that Intel has "lost its way."

An interview with Gates by the Associated Press notes how the billionaire has a soft spot for Intel. The publication suggesting that his career might have gone down a different path had Team Blue not created the first commercial microprocessor, the Intel 4004 in 1971. It led to more advanced chips that powered personal computers, resulting in the need for software for these PCs.

While Microsoft has been on the rise since Satya Nadella became CEO in 2014, Intel has endured its most difficult period in decades. There were delays transitioning from the 14nm to 10nm process, followed by the delay of 7nm. Intel has also constantly lost market share to AMD, faced Apple dropping the company in favor of its own silicon, dealt with security vulnerabilities, struggled with Raptor Lake issues, and lost ground to chip rivals. Financial troubles have also hurt the company, culminating in the ousting of CEO Pat Gelsinger last year.

"I am stunned that Intel basically lost its way," Gates said. He added that Intel co-founder Gordon Moore "always kept Intel at the state of the art. And now they are kind of behind in terms of chip design and they are kind of behind in chip fabrication."

Nvidia, TSMC, and Qualcomm are all ahead of Intel in various areas of chip manufacturing and design, and catching up isn't going to be easy, if not impossible.

Gates also highlights how Intel essentially missed the AI chip revolution, though he did have praise for former CEO Pat Gelsinger.

"I thought Pat Gelsinger was very brave to say, 'No, I am going to fix the design side, I am going to fix the fab side.' I was hoping for his sake, for the country's sake that he would be successful. I hope Intel recovers, but it looks pretty tough for them at this stage."

Related: Intel's takeover dilemma: A Gordian knot of funding and politics

Intel has also been falling behind AMD in the consumer CPU market. Team Red is dominating the Amazon.com processor sales chart while continuing to do well abroad. Intel's only solace could be that more Steam survey participants (63%) still use its CPUs.

There have been rumors that Intel could be bought out – Broadcom looked like a potential buyer for a while – but funding the company's fabs will require tens of billions of dollars and take years to get back on track, making it a less appealing proposition. Given the amount of money the US government has poured into them, shutting down its fabs isn't an option for Intel.

Permalink to story:

 
I see this as one more billionaire starting to publicize a bunch of their opinions that no one asked for.

Anytime a billionaire offers a statement, the average person would be well served to ask themselves "what are they gaining by sharing this?"

100% they asked themselves that question first.
 
Intel stopped being an industry leader when they started acting like a monopoly.

The OEM bribes are still flowing and is the only reason why Intel is able to maintain their share in mobile. The day the money will stop flowing is the day AMD will eat their share.

I hope it is sooner than later. They deserve no sympathy.
 
AMD has been consistently offering better and cheaper CPUs over the past decade, yet Intel maintains a dominant market share. This isn't a failure; it's a testament to their success. After 10 years of being the second in performance but first in cost CPU provider, Intel should have a market share of 1-5%, not 60-70%. For comparison, AMD holds about 15% of the GPU market because has less performance than Nvidia and despite being cheaper than Nvidia.

Intel faces just three challenges. First, their fabrication plants (fabs) are lagging behind TSMC's in terms of node advancement and yield. To mask this, they employ unconventional techniques and introduce asymmetrical core designs that don't align with the symmetric thread management expected by the operating system. They opt for less silicon engaging ("efficient" cores) to reduce overheating, but this is forced due to the less advanced fabrication nodes they have.

Second, Intel's CPUs are affected by the meltdown issue, unlike AMD's, which leads to performance degradation only on their chips from the need for mitigation strategies through code.

Third, Intel does not implement 3D caching, which could enhance performance and efficiency. To reach in the RAM for data cost 300 cpu cycles instead of 100 cycle for L3 cache 20 cycles for L2 and 3 cycles for L1.

If Intel addresses these three issues—improving fab node advancement, aligning core designs with operating system expectations and adopting 3D caching they could reclaim a significant portion of the CPU market, potentially reaching again the 90%.

However, even without these improvements, Intel continues to sell more expensive and less performant chips and consumers still purchase them. This is puzzling because, unlike operating systems where users are often forced to use the OS due to application compatibility, CPUs are hardware components that users can freely replace. Despite Intel's CPUs being less efficient and more expensive, they remain popular because most users primary use cases are basic tasks like document editing, web browsing and occasional gaming, which today the CPUs can handle adequately.
The most significant issue isn't that Intel doesn't produce competitive CPUs; it's that consumers continue to choose Intel's products over better alternatives.
After all is a responsibility for the people in Intel to put the maximum effort to keep that great company in good shape even if their competition doesn't penetrate the market.
 
Intel faces just three challenges. First, their fabrication plants (fabs) are lagging behind TSMC's in terms of node advancement and yield.
Yep, and this is their biggest problem imo. It's what kept them ahead of the curve for a long time, allowing them to offer what AMD simply couldn't. However right as they started having trouble with keeping this advantage AMD struck gold with the Zen architecture.

To mask this, they employ unconventional techniques and introduce asymmetrical core designs that don't align with the symmetric thread management expected by the operating system. They opt for less silicon engaging ("efficient" cores) to reduce overheating, but this is forced due to the less advanced fabrication nodes they have.
"Necessity is the mother of invention"
That might work out, mobile chips have done this for a long time with all kinds of configurations - even combining 3 different types of cores. Mobile chips however tend to deal with the Linux kernel (ignoring Apple/iOS here) which seems to adept to changes much more easily than Windows and its kernel.
A lot of AMDs chiplet designs with core complex latency issues and Intels problems due to switching between the two different types of cores are much more pronounced under Windows.

It isn't necessarily a bad step, just one that needs a bit of time to smooth things out (I think they mostly have now?). [Personally I still think that Microsoft once their marketshare shrinks enough will move to the Linux kernel at some point as well so they can stop having to develop their own]

Second, Intel's CPUs are affected by the meltdown issue, unlike AMD's, which leads to performance degradation only on their chips from the need for mitigation strategies through code.
This problem seems two-fold. One was them pushing the silicon/architecture too hard to stay competitive with AMD (with the insane power usage as a result). Second was them having purity problems in some of their fabs.

Supposedly their fabs are okay now resolving the issue, but man did that ever do damage to their reputation and cost them some important enterprise customers. The type of customer that once they made the switch to AMD is very unlikely to ever come back unless AMD pulls a stunt like that themselves.

We'll have to wait for a second generation of the arrowlake architecture to see if they're still forced to push the silicon so hard but going by the first gen even though performance was underwhelming - effiency wise it was a big step that was sorely needed.

Third, Intel does not implement 3D caching, which could enhance performance and efficiency. To reach in the RAM for data cost 300 cpu cycles instead of 100 cycle for L3 cache 20 cycles for L2 and 3 cycles for L1.
I do wonder if they have internally tested this (actual silicon or just simulations) and didn't see any worth in pursuing it. It could be that their architecture just scales a lot worse with a greatly increased cache compared to AMDs. Or it could be that they don't want to rely on it because they plan on bringing production back in-house and they can't do it themselves (don't have the technology and/or licensing it from TSMC is not an option or too costly).
The applications for it do seem limited as aside from gaming (at least with AMDs architecture) other tasks don't seem to see much if any benefit from it.

However, even without these improvements, Intel continues to sell more expensive and less performant chips and consumers still purchase them. This is puzzling because, unlike operating systems where users are often forced to use the OS due to application compatibility, CPUs are hardware components that users can freely replace. Despite Intel's CPUs being less efficient and more expensive, they remain popular because most users primary use cases are basic tasks like document editing, web browsing and occasional gaming, which today the CPUs can handle adequately.
Most users (read: people that use a PC but don't visit sites such as this one) just buy a PC, from a brand they know with brands in it that they know. Intel is pretty well known (not quite Coca Cola but they're fairly high up there). As long as that PC does what they expect from it they'll keep visiting positively of those brands. If it doesn't do what they expect it to do... well, they probably bought an ARM-based PC and will from there on avoid the name Qualcomm in their PC purchases ;)
 
What's with all the sarcasm against Bill, lol?
If anything I wish he'd still be leading Microsoft, considering what a mess windows 11 is.
What's with all the sarcasm against Bill, lol?
If anything I wish he'd still be leading Microsoft, considering what a mess windows 11 is.
How true, he needs to get back over to MS and sort out the mess that is Win 11 component requirements and Win11 updates
 
AMD has been consistently offering better and cheaper CPUs over the past decade, yet Intel maintains a dominant market share. This isn't a failure; it's a testament to their success. After 10 years of being the second in performance but first in cost CPU provider, Intel should have a market share of 1-5%, not 60-70%. For comparison, AMD holds about 15% of the GPU market because has less performance than Nvidia and despite being cheaper than Nvidia.

Intel faces just three challenges. First, their fabrication plants (fabs) are lagging behind TSMC's in terms of node advancement and yield. To mask this, they employ unconventional techniques and introduce asymmetrical core designs that don't align with the symmetric thread management expected by the operating system. They opt for less silicon engaging ("efficient" cores) to reduce overheating, but this is forced due to the less advanced fabrication nodes they have.

Second, Intel's CPUs are affected by the meltdown issue, unlike AMD's, which leads to performance degradation only on their chips from the need for mitigation strategies through code.

Third, Intel does not implement 3D caching, which could enhance performance and efficiency. To reach in the RAM for data cost 300 cpu cycles instead of 100 cycle for L3 cache 20 cycles for L2 and 3 cycles for L1.

If Intel addresses these three issues—improving fab node advancement, aligning core designs with operating system expectations and adopting 3D caching they could reclaim a significant portion of the CPU market, potentially reaching again the 90%.

However, even without these improvements, Intel continues to sell more expensive and less performant chips and consumers still purchase them. This is puzzling because, unlike operating systems where users are often forced to use the OS due to application compatibility, CPUs are hardware components that users can freely replace. Despite Intel's CPUs being less efficient and more expensive, they remain popular because most users primary use cases are basic tasks like document editing, web browsing and occasional gaming, which today the CPUs can handle adequately.
The most significant issue isn't that Intel doesn't produce competitive CPUs; it's that consumers continue to choose Intel's products over better alternatives.
After all is a responsibility for the people in Intel to put the maximum effort to keep that great company in good shape even if their competition doesn't penetrate the market.
I hope Intel sees the wisdom in your post they should really consider hiring you for at least some consulting.
 
Back