»Bringing the Geek back« is written on Gelsinger’s shirt – according to him, these are the things that the company has been missing for a long time.
Pat Gelsinger has been CEO at Intel for around 18 months. And since the manager’s return to the semiconductor giant, a lot has app arently happened.
Im Interview with Nilay Patel from US Podcast Decoder the Intel boss speaks openly about mistakes that have been made in the past and the mission he is now pursuing: Intels lost decade
leave behind and rebuild the company from the ground up.
The CEO addresses four major mistakes that have been made at Intel in recent years and that have collectively led the company down the wrong path.
Some links included on this page are affiliate links. Depending on the provider, GameStar receives a small commission for purchases made via these links without affecting the price.
More info.
A lack of engineering culture at Intel
According to Gelsinger, Intel had lost its core corporate culture over the years. It was methodology, process and a general engineering culture that made the company great.
Many fundamental values such as discipline, decisions based on hard data and an orientation towards the final performance would have suffered as a result.
A condition that, according to Gelsinger, should no longer last. Bringing the Geek back
is his motto that he had binary and ASCII code printed on his shirt for the recent Intel InnovatiON. You can watch the highlights of the event on day one in the following video:
15:47
Intel Arc, Raptor Lake and Co.: These were the highlights of Intel InnovatiON
There were also the wrong people at the management level when he started. There would have been many good leaders there, but they were in jobs for which they not qualified
would have been. In the past few months, Gelsinger had therefore replaced around 70 percent of Intel’s management level.
The end of the tick-tock model
A principle that had been in use at Intel for twenty years was also abolished in Gelsinger’s absence: the tick-tock model. This is how Intel described its approach to the development of new processor designs.
With each new generation of processors, either the manufacturing technology (tick) or the microarchitecture (tack) was further developed. But not both at once.
In this way, constant jumps in the products could be guaranteed with little risk. In the interview, the CEO was unable to say why the tick-tock model was being ended in Gelsinger’s absence: Why did we stop? That was extremely successful. let’s do it again Let’s put it back in.
more on the subject
Raptor Lake: This is how Intel wants to get the CPU crown back
We risked too much – and failed
In general, risk management seems to have been a big sore point at Intel over the last few years. You wanted too much at once and got paid for it.
Some of these past mistakes still plague Intel today, says Gelsinger. The server processors, alias Sapphire Rapids, have been in the works for five years because they wanted to change too many things at once: DDR5, PCI Express 5.0, a new standard for the CPU device connection (CXL).
His solution? We need to flush the pipes of these projects. We must complete them, complete them and restore the culture of quality.
At the same time, however, important bets on the future were backed by the wrong horse. He cites EUV (Extreme Ultraviolet Lithography) as an example. The technology would have allowed further miniaturization of the microarchitecture, but the company wanted to avoid EUV for the Intel 10 architecture.
Now the competitor TSMC, where Apple and AMD have their semiconductors manufactured, has a big lead when it comes to using EUV – although Intel, according to Gelsinger, was actually responsible for the fundamental development of the technology.
By the way: TSMC could also have problems in the future. But they are more due to the energy requirements of the company:
The world’s most important chip manufacturer could soon have a problem – and so can we
The end of the song: lost trust
All of this has led to a loss of customer confidence over the years, says Gelsinger. It could not have offered the expected quality, quantity, and performance that customers have come to expect from Intel’s products.
Now we want to regain that trust. And according to Gelsinger, you already get better feedback from big customers, even if you haven’t reached the end of the road yet.
In order to iron out the mistakes of the last decade, Gelsinger wants to think long-term. The strategy is not designed for a quarter or a year. Instead, they are already planning for the coming decade. During this time, they want to aggressively rely on EUV to advance microarchitectures and get rid of legacy assets – even if Intel’s shareholders have to accept short-term slumps in share value as a result.
Intel should also show staying power with its new graphics cards. While the first Arc cards aren’t the catastrophe that many predicted, as our test overview below shows, there is hope for generations to come.
Tests of the new Intel Arc graphics cards reveal a surprisingly complicated picture
Do you see it like the Intel boss that the last few years for Intel a lost decade
were? And can the semiconductor giant soar to a new heyday? Or is the company running out of breath? Tell us your opinion in the comments!
Table of Contents
