Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/intel-considers-drastic-change-to-catch-ai-rivals.20921/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Intel considers drastic change to catch AI rivals

XYang2023 said:
www.irishtimes.com

Intel considers drastic change to catch AI rivals

Senior departures and plunging share price hit chief executive Pat Gelsinger’s plans for company turnaround
www.irishtimes.com www.irishtimes.com

What drastic changes?
Click to expand...
Some interesting paragraphs:

"Intel chief executive Pat Gelsinger received an unusually direct demand from the company board in the spring of last year: pay closer attention to its strategy on artificial intelligence.

According to three people familiar with the directive, the board expressed concern it would miss out on a new multibillion-dollar market to create chips that power generative AI which had emerged in the months after OpenAI launched ChatGPT.

Gelsinger’s response was to set up an AI Acceleration office, charged with co-ordinating its AI plans across multiple business segments. The office would be led by Srinivas Lingam, transferred to California from his post at its AI product group in India."
----
"Instead, attention is focused on Gelsinger’s relationship with his board. In August, a key board member charged with overseeing its crucial chip manufacturing strategy resigned. While the AI Acceleration office continues its work, Lingam moved to another position at Intel earlier this year."
----
"Recently departed staff said that, while they agreed with Gelsinger’s strategy, the company is fumbling its execution. They described frustration at a sprawling internal bureaucracy, with a back and forth over lay-offs sapping morale."
----
"The board has pushed for more oversight of Intel’s strategy. Last year, Lip-Bu Tan, former chief executive of chip design software company Cadence, who was appointed to the board in 2022, was given additional responsibilities to guide the company’s critical foundry business. The role came with an additional restricted stock award of around $1 million. In August, Tan departed, fuelling further anxiety about the top of the company. Tan did not respond to requests for comment."
 
Intel needs to focus on CPUs and see if they can get GPUs to work. Intel is very good at CPUs
Pat chose to focus on foundry, Intels worst and biggest money losing area. The obvious thing happened.

Intel is too slow for GPUs. see if you can speed development by 2x and the GPU market will come. Hire 5 people from nvidia and it will become VERY clear what intel needs to change. Intel people who left and went to nvidia are very clear on what needs to change.

My fear is that Intel is like an football team that goes 8-8 Four years in a row. They say: "we need to make big changes and not tolerate mediocrity" . 3 years later they are 3-13. Congratulations, you are no longer mediocre.

Just be great at CPU, leading architecture direction, learn how to do GPU better and faster. stock price will hit 50.

just an opinion
 
Last edited:
Maximus said:
"According to three people familiar with the [board] directive [to "pay closer attention to its strategy on artificial intelligence"], the board expressed concern it would miss out on a new multibillion-dollar market to create chips that power generative AI which had emerged in the months after OpenAI launched ChatGPT."
Click to expand...

MKWVentures said:
Intel is too slow for GPUs. see if you can speed development by 2x and the GPU market will come. Hire 5 people from nvidia and it will become VERY clear what intel needs to change. Intel people who left and went to nvidia are very clear on what needs to change.
Click to expand...
Is Intel culturally capable of listening?

Intel has already missed the AI boat, and the problem starts at the top where Pat Gelsinger insufficiently appreciated the ecosystem required for a foundry. That's in his domain as a designer, expecting him to do it for the programmers and system administers of GPUs is unrealistic.

Compare AMD's and Intel's consistent failures with Jensen Huang spending two decades and billions on an ecosystem to let GPUs do anything including making masks at TSMC.

AMD's Lisa Su claims to be proud of her software ignorance. but she has gotten in the Twitter trenches over her company's catastrophic GPU's failures when they go off the happy path and require a system reboot. I can't see what Intel could offer to those already at Nvidia.

For Intel board members who might understand ecosystems, maybe James Goetz, Andrea Goldsmith, Tsu-Jae King Liu, and Dion Weisler, while Alyssa Henry defintely should from her time at AWS.
 
MKWVentures said:
Just be great at CPU, leading architecture direction, learn how to do GPU better and faster. stock price will hit 50.
Click to expand...
A terrible idea. The personal computer CPU market keeps shrinking. After all those computer sales during the lockdowns you will also probably not see a lot of sales over the next years. In the meantime the capex cost to make a fab keeps increasing. Intel needs to have a product to keep those fabs operating. I think GPUs are a decent fit as an addition to CPUs.

Intel needs to execute on their server processors to claw back some of what they lost in terms of datacenter revenue. They need to move server processors to more modern processes, use chiplets to reduce issues with yield, and work out some way to optimize CPUs to work on AI workloads. Some kind of AVX extension for example.

The GPUs will help move laptop product, and might be used for datacenter acceleration eventually. But for that Intel needs to do massive investments in software to compete against NVIDIA's CUDA. Something NVIDIA's competitors have thus far failed at. The antitrust lawsuit against NVIDIA might provide an opening for Intel and AMD here. They could demand NVIDIA open up CUDA as an industry standard for example.
 
Last edited:
samwilde said:
A terrible idea. The personal computer CPU market keeps shrinking. After all those computer sales during the lockdowns you will also probably not see a lot of sales over the next years. In the meantime the capex cost to make a fab keeps increasing. Intel needs to have a product to keep those fabs operating. I think GPUs are a decent fit as an addition to CPUs.
Click to expand...
I would argue it's only growing over time, just slowly, people need at least 2 computer per person, one for work and one for personal. The AI story continues to unfold, making professional jobs easier. But most jobs cannot be done in mobile. People need PCs/mac for a better screen experience. I would argue the expansion into console is one extension story of personal computing. There are a lot of applications to be developed not only in PC, console, appliances, AR/VR headset, car.

With lunar lake, Intel demonstrated they can deliver in low power setting. So not only Qualcomm is trying to eat Intel's lunch. Intel also trying to eat theirs.

samwilde said:
Intel needs to execute on their server processors to claw back some of what they lost in terms of datacenter revenue. They need to move server processors to more modern processes, use chiplets to reduce issues with yield, and work out some way to optimize CPUs to work on AI workloads. Some kind of AVX extension for example.
Click to expand...

samwilde said:
The GPUs will help move laptop product, and might be used for datacenter acceleration eventually. But for that Intel needs to do massive investments in software to compete against NVIDIA's CUDA. Something NVIDIA's competitors have thus far failed at. The antitrust lawsuit against NVIDIA might provide an opening for Intel and AMD here. They could demand NVIDIA open up CUDA as an industry standard for example.
Click to expand...

I agree with what's being said here. I think unless Nvidia stumble in a stable market environment, Intel/AMD may have a chance. But now it is too pre-mature.
 
delong.height said:
I would argue it's only growing over time, just slowly, people need at least 2 computer per person, one for work and one for personal. The AI story continues to unfold, making professional jobs easier. But most jobs cannot be done in mobile. People need PCs/mac for a better screen experience. I would argue the expansion into console is one extension story of personal computing. There are a lot of applications to be developed not only in PC, console, appliances, AR/VR headset, car.

With lunar lake, Intel demonstrated they can deliver in low power setting. So not only Qualcomm is trying to eat Intel's lunch. Intel also trying to eat theirs.





I agree with what's being said here. I think unless Nvidia stumble in a stable market environment, Intel/AMD may have a chance. But now it is too pre-mature.
Click to expand...
Car manufacturer also want a third alternative. Intel got mobileye, and they should be able to compete with both Nvidia and Qualcomm in this regards
 
You must log in or register to reply here.
Back
Top