How Intel lost the crown in the processor race

Intel has been the king of the processor market since its inception, and many thought it would always be so. 50 years of absolute dominance and superiority, recognition and success, ended with one decision of a man who was recently considered a saint. One decision that set Intel back a decade and destroyed all financial prospects. It was a mistake that has sent Intel shares more than half over the past 9 months. It was a mistake worth 50 years.

Today, Intel costs almost three times less than its eternal competitor, AMD. And there was… just one reason.

And her name is Pat Gelsinger.

Start. How Zen 3 broke the back of Titan

The circumstances that led Intel to today's sad outcome began long before recent events. In 2020, AMD introduced another line of Ryzen processors, this time based on the Zen 3 architecture. This was the Ryzen 5000 family, designed to develop the potential of the innovative chiplet technology presented in Zen 2, as well as improve the weak points – performance per core, and the need for using power plans due to problems with the Windows 10 scheduler. Optimistic expectations from the new architecture were not out of the ordinary, but it was at that moment that a turning point occurred, which became the curse of Intel for all subsequent years.

The stumbling block was… energy efficiency.

At first glance, this may sound strange, because at that time Intel had not yet suffered from serious power problems, and company director Bob Swan recommended Look less often at benchmarks, the success of AMD processors in which was difficult not to notice. However, Intel's abundant investments in the UserBenchmark project helped to somewhat cool down the interest of the mass public in red processors – albeit not for long.

But Intel actually had the problem of temperatures and power supply even then – the i7-10900k, released in the same 2020, operated at high frequencies above 5 GHzbut at the same time it consumed significantly more than the rated 125 W, heating up dangerously close to the maximum 100 degrees specified in the manufacturer’s specifications.

The architectural limits of Comet Lake had actually been reached, but unfortunately for Intel, the upward movement was just beginning for AMD.

Sales growth EPYC server processors broke all records in 2020, and only accelerated in 2021 when Milan processors with cores based on the Zen 3 architecture entered the market in March.

Intel answered aggressive discounts on its Cascade Lake family of Xeon solutions, but it was obvious that without radical changes the company would continue to lose ground. It was then that the inevitable happened – Bob Swan resigned, and on February 1, 2021, Pat Gelsinger became the head of Intel.

Everyone was expecting the beginning of a new era. And a new era has truly begun.

Sing and dirty copper

Pat Gelsinger is not a financier or a manager. Intel's board of directors, clearly inspired by the leadership example of Lisa Su (who pulled AMD out of the bottom with a capitalization of just over a billion), chose a person with equally recognized talents.

The honored engineer, one of the main developers of the legendary 80386 processor (which even has his initials engraved on it), the leading developer of the Intel Core architecture, which once made the company an industry leader after the difficult 2000s, found himself in the chair of the head of Intel and immediately got to work.

Among the global tasks of the new director of Intel was, for example, solving the problem of the company's technological lag in production technologies. Gelsinger promised to create up to 20 thousand new jobs in the US with the deployment of the Intel production complex on home soil – plants in Arizona, New Mexico, Ohio and Oregon, which received Biden's support and became a kind of advertisement for the “new era of Intel”, stalled at the stage bureaucracy – and according to the latest estimates, should start no earlier than 2026.

Meanwhile, Intel's internal management fell into chaos. While various teams were working on designs for future core architectures, AMD unexpectedly introduced the 5800X3D, a processor whose distinctive feature was the use of a vertical die layout, where large blocks of L3 cache were located on top of the core processing units, communicating with them through copper insulator conductors.

It was the development of copper insulators in his thesis that once described Lisa Su, future director of AMD.

The processor instantly destroyed the Blues' gaming advantage.

A fast and large cache solved all latency problems and provided performance comparable to the i9-12900k in many processor-dependent games. But the most humiliating thing for Inrel here was that power consumption with such a jump in performance practically did not increase against the background of efficient Zen 3 processors – and the new processor from AMD in stress tests at maximum load consumed 2 times smaller than Intel's flagship.

This was the moment when everyone realized that the Blues had a problem.

All hope was for a completely new, innovative architecture, for which Intel invited the legendary Jim Keller to create a new breakthrough base for future processors.

The innovative core, which the media managed to nickname Royal Core, was never developed – due to the fact that Jim Keller and Pat Gelsinger couldn't find a common language while working on the project.

The breakthrough architecture, on which Intel's fate largely depended, went into the trash. What was left? Test the limits of the current architecture for strength.

And this was precisely the main problem. While AMD was tidying up the AM5 platform, mentally preparing users for a costly upgrade, Raptor Lake, Intel's 13th and 14th generation processors, were fast and productive, but achieved their goals solely by increasing frequencies, which in addition to two types of cores (P- core and E-core – a productive and energy-efficient core, respectively) heated the crystal to the limit when consumption limits were removed.

The appetites of processors striving for the 6 GHz mark in frequencies sometimes approached the 400 W mark, which, against the background of cold and powerful Ryzens, looked like some kind of absurd. The release of the 7800X3D, which combined the energy efficiency of Zen 4 with an extremely high-speed cache buffer, was the last straw for the “blues” – AMD took away Intel’s gaming advantage so brazenly that even the 14900k did not look like an equivalent solution. It was a dead end in every way.

But his discussions would have remained conflicts between fans on the Internet, if not for one small incident at the Intel plant.

Sometime in 2022, a production line was contaminated at the Intel plant. Because of oxidation Copper contact pads under the processor chip eventually deteriorated electrical conductivity and the chip completely failed.

This problem appeared to be very local at first glance – and for a long time it was believed that the number of processors affected by it was small. At the same time, taking into account the dating, it was believed that the problem affected only the production of 13th generation processors, and did not affect Raptor Lake Refresh, released in October 2023.

And it was at the moment when information about the possible contamination of the line became public that Pat Gelsinger made his main mistake. A mistake that cost Intel hundreds of billions of dollars.

He said nothing.

The first difficulties with Raptor Lake processors began to appear at the end of autumn 2023 – then there were isolated reports on forumswhere people complained about crashes, blue screens, freezes and other things that PC gamers and power users are accustomed to associate with typical hardware problems – power failures, insufficient cooling, incorrect memory configuration. At first glance, everything was exactly the same here, but over time, messages from concerned owners of Raptor Lake (and then Raptor Lake Refresh, 14th generation) became more and more numerous. The symptoms were similar for everyone, but deaths began to be added to the problems of instability. Expanding the geography of problems affected and nVidia graphics drivers, after which by the spring of 2024 the number of statements from affected users no longer fit into the neat list of “losers.”

The moment of truth has come. The head of Intel could solve all the problems that arose with one precise, clear statement.

April 29, 2024 Intel published official statement.

“Intel has discovered that the problem may be caused by high voltage and high heat,” the company said. “This occurs when processors operate outside of Intel's recommended specifications <...> While the root of the problem remains unclear, the majority of error reports come from owners of motherboards with unlocked/overclocking support.”

Yes, you didn't think so. Intel was blamed for the problems on motherboard manufacturers.

These statements were instantly circulated by all media. Users rejoiced – after all, it was enough to adjust the settings, and everything would immediately return to normal, right? As time has shown, no.

But the company did not voice the real cause of the problems. Instead, a few weeks later Intel announced that problem appeared… due to errors in microcodeand the update was, of course, “already in the works.”

But as Intel clarified, all this was not the cause of instability. The reason turned out to be fatal.

The moment of truth. The market changes sides

The most vulnerable points for Intel have always been two market segments – laptops and the server market. Both the first and the second were undividedly owned by Intel for many years, bringing the company tens of billions of dollars a year, but the emergence of Ryzen began to muddy the waters not only in the notebook direction, but also in the critical server segment. With the increasing performance of the Zen architecture, more and more companies began to show interest in the prospects of purchasing EPYC-based servers to replace aging Intel solutions, and therefore the “blue” initially sharply reduced prices on Xeons, and then focused on advertising new generations of processors – however, this did not become any problem for AMD.

The beginning of difficulties with the stability of processors based on Raptor Lake did not leave large customers of the chip giant indifferent – many began to wonder whether the problem of stability would affect server solutions? And if it does, how extensive will the damage be?

It was questions like these that forced the corporation to release its long-awaited statement. In July 2024, Intel finally admitted that the main stability issue was due to contamination on the production line. The same thing that the company stubbornly remained silent about for almost two years.

Investors reacted instantly. Intel shares lost a third of their value in a matter of days, and given the gradual decline throughout 2024, the legendary company began to cost almost 3 times less than AMD. The same AMD, which just recently was less than half the capitalization of Intel.

Many large companies, in light of recent events, began moving to the latest EPYC processors, and Intel focused on trying to minimize the public damage to its reputation – but did not admit that up to 8 million processors of two generations could be at risk of damage due to a manufacturing error I made up my mind.

Intel and the future. Where is the great blue going?

The outstanding engineer Pat Gelsinger turned out to be the most incompetent company leader in at least the last 25 years. Instead of creating that very “Royal Core” under the leadership of Jim Keller and ushering in a new era, not inferior to the success of Intel Core, he quarreled with the legendary engineer, canceled and closed several key architectural projects (and “reorganized” the GPU division), in ultimately making a key decision that took away the crown from Intel, which, it would seem, always belonged to the corporation. A cascade of problems that arose, strange management decisions and incompetent work with partners ultimately led the company to the worst period in more than half a century of active work.

But the most important thing is not financial losses or reasons for fighting on the Internet. The most important thing is that the key giants of the IT industry (from Microsoft to Amazon) have seen with their own eyes what unconditional trust in Intel is worth. For all 2 years, while the corporation's management (and Pat Gelsinger personally) knew about the production problems, no one rushed to notify partners about the possible risks for Xeon processors, and when the time came to present evidence, the company did not have it in its hands. The losses of key clients represented by megacorporations create unprecedented financial risks for Intel, because the server segment is tens of billions of dollars a year, which AMD has been slowly biting off since 2018. The previous 8% of the market turned into 24%, which means that a quarter of the insanely profitable market segment has already gone to a competitor. A similar story is unfolding in the desktop segment – during the time of the great “Fufyks” the share of AMD processors did not exceed 8%, but after the advent of Ryzen the situation began to gradually change, and today the previous 8% has turned into 36.7% – more than a third of the fleet of gaming machines on Steam are built on processors from AMD.

But sincere Intel fans, expecting a beautiful comeback, are confused. They do not see on the horizon the long-awaited revolution that was talked about so much a few years ago. Foveros 3D layout project (about which talked back in 2018!!!) according to Gelsinger, it is about to change the traditional layout in the production of processors, but in reality it may still be as far from being realized as before. Rebranding Intel Core processors, which will now be called Ultra 9, Ultra 7 and Ultra 5, have already managed to discourage fans with the lack of hypertrading and strange gaming performance. Trying to get away from the associations with the names 14700k and 14900k flashing in the news and on YouTube seems like an understandable PR move, but we still don’t know what was the root cause of the contamination of production lines – and we can’t be sure that it won’t happen again.

What's the result? In light of recent events, when Intel and AMD recently announced the creation of a “consortium of x86 leaders,” essentially declaring war on the upcoming ARM initiatives and individual processor development projects (one of which, for example, is engaged Google), it becomes clear that there is no longer any solid ground under the feet of the once great corporation. Just a few years ago, it was unimaginable that Intel would stoop to become a competitor and enter into some kind of strategic partnership. But time passes and everything changes. In the end, you can always correct a damaged public image or win the race with black PR, but unfortunately, it is impossible to regain the lost trust of large companies that saw with their own eyes what the notorious Intel Inside is really worth.