Intel’s manufacturing crisis puts company at a crossroads

Intel's D1X research factory in Hillsboro, where the company develops and makes each new generation of microprocessor. (Credit: Intel Corporation)

Intel’s Oregon scientists have a reputation for working magic.

The company crafts each new generation of its microprocessor technology at its Ronler Acres campus in Hillsboro, where researchers working on scales of just a few billionths of a meter create new technologies that make our computers faster, more powerful and efficient.

Skeptics have warned for years that this cannot go on forever, that as chip features approach the atomic level, Intel will eventually hit a wall that slows progress – or stops it altogether.

Intel has made sport of defying the doubters – deploying new materials, remaking the transistor and performing other tricks to enable continued innovation.

Now, something has gone terribly wrong.

Three generations of Intel chip technology have all been hobbled by delays and setbacks. Its current 10-nanometer chips were four years late when they hit the market last winter, an eternity in the fast-paced tech sector. Then the company shocked investors last month by disclosing that its forthcoming 7nm chips, previously due in 2021, are already running a year behind.

Worse yet, Intel didn’t seem to see its problems coming. As recently as a few months ago, Intel was assuring Wall Street and computer manufacturers that the company had learned from its mistakes on 10nm and that the new, 7nm chips were solidly on track.

The lapse on 7nm has thrown Intel into crisis. The company shed $42 billion in market value in a single day after it announced its manufacturing problem, despite reporting blockbuster financial results.

The following week, Intel parted with its chief engineer and its chief technology officer announced his retirement as the company shook up its manufacturing organization.

Intel, long the world’s most advanced chip producer, now clearly trails rival Taiwan Semiconductor Manufacturing Co., which has continued delivering successive chip upgrades on schedule.

And in a humiliating development, Intel said it has “contingency plans” to outsource its own production to contractors like TSMC if Intel proves incapable of manufacturing its most advanced chips itself.

The question now is whether Intel can play catch-up and restore its leadership position, or if it will have to resort to some fallback position that cedes the most advanced manufacturing to its rivals – perhaps even selling off some of its factories, which the chip industry calls fabs.

Intel shows no sign of changing course, telling employees this month that owning its own fabs gives the company “the ability to control our own destiny.”

Technology and investment analysts, though, are split. Some assert that it will take years for Intel to restore its leadership, if it ever does, and arguing it’s time for a dramatic change. Others note that Intel’s sales are booming, even during the coronavirus pandemic, adding that Intel’s research in advanced materials and transistor design give the company the opportunity to rebound.

Whatever happens will have profound implications in Oregon, where Intel is the state’s largest corporate employer. Though Intel’s headquarters are in Silicon Valley, its 21,000 Washington County employees lead the company’s research and manufacturing and are responsible for blazing a path for its future.

“I think Intel can do it. They’ve certainly got the money and the right people. They’ve got some of the smartest people in the world up in Oregon,” said Dan Hutcheson, a veteran industry analyst and head of VLSI Research. “It’s really a matter of, can they work together and pull this thing off, and start making these moves quickly?”

‘Golden age’ is over

Manufacturing computer chips is an extraordinarily complex process, using light to imprint microscopic patterns and then etching away at the surrounding materials. It’s a laborious, exacting process repeated in multiple steps for each chip, when even the tiniest error will wreck the entire product.

Defects are common, but usually manageable. Intel and other chipmakers simply discard the chips that don’t meet their performance specifications. The functioning chips are valuable enough that Intel frequently enjoys gross profit margins around 60%.

What’s happened recently is that Intel hasn’t been able to control the rate of defects in its newest chips. That means there are so many flaws in the early prototypes that it takes much longer than usual to refine the manufacturing process and reduce the number of defects to an acceptable level.

To some degree, this appears to be the inevitable consequence of the minute scales at which Intel operates. For all of its five decades the company’s technology has been guided by Moore’s Law, a maxim coined by Intel co-founder Gordon Moore, which states that computer performance will improve at an exponential rate, at regular intervals.

Those exponential improvements enabled generations of breakthroughs in computing power at successively lower cost, producing breakthroughs from the desktop PC to laptop to smartphones and data centers. Intel led this advancement by continually shrinking the features on its chips.

It’s always been obvious that at some point Moore’s Law would run headlong into the laws of physics – that there would come a moment at which features simply couldn’t get any smaller. Intel appears to be near that point.

In a technology presentation this past week, Intel acknowledged the “golden age” of Moore’s Law is over, noting that the rate of improvement from each new generation of chip has been declining for years.

“The era of getting massive performance boost from simply shrinking transistor features is behind us,” Intel fellow Ruth Brain said Thursday.

Intel has compensated by making improvements within each new generation of processors, to wring out better performance without shrinking the features. What’s striking about Intel’s troubles with its 10nm and 7nm chips, though, is that the company appears to have been caught off guard by its failures.

That indicates this is more than a technical challenge. It suggests internal organizational problems.

“I think that’s probably the most reasonable explanation to why they’re struggling, is they just haven’t been able to pull it together as a team,” Hutcheson said.

Intel endured wrenching layoffs in 2015 and 2016, the biggest in the company’s history, which shook the company’s culture and demolished morale. Then came what Hutcheson describes as Intel’s “CEO mess.”

Former chief executive Brian Krzanich, who engineered the layoffs, quit abruptly in 2018 after the company uncovered “a past consensual relationship with an Intel employee” in violation of corporate policy. It took Intel seven months to settle on a replacement CEO, ultimately choosing chief financial officer Bob Swan.

Meanwhile, Intel has reorganized its manufacturing group several times – including once in June and again in July, after Intel disclosed the crisis with its 7nm chips.

Renowned chip architect Jim Keller unexpectedly quit in June and chief technology officer Mike Mayberry said last month he will retire at the end of the year. Intel parted with chief engineer Venkata “Murthy” Renduchintala, who earned an astonishing $27 million last year, immediately after the company disclosed the trouble with its new processors.

A symbol of decline?

Through it all, Intel’s sales have kept climbing. Revenues were up more than 20% in the first half of the year (though Intel projects a considerably slower second half) and Intel dominates the lucrative data center market.

Intel’s factories are full -- in fact, the company suffered a severe shortage of production capacity last year and is spending billions of dollars to build or expand leading-edge fabs in Oregon, Arizona and Israel.

And Intel remains a technical innovator. On Thursday, it unveiled new features in a forthcoming upgrade to its 10nm processor, called Tiger Lake, which the company said provides a 20% boost in performance – as big a performance boost as if it had released a whole new processor.

What concerns Wall Street is that chronic manufacturing setbacks could open the door to competitors.

Apple dumped Intel in June in favor of designing its own chips for its Mac computers based on designs from Intel rival ARM Holdings. Other PC manufacturers are sure to follow, and in time Intel could lose a significant share of the data center market, too, if it can’t hold a competitive edge.

Bloomberg opinion columnist Noah Smith wrote last month that Intel is a “symbol of big tech’s decline” and said the company simply lost its way.

Intel has acknowledged it was too aggressive with the 10 nanometer chips that came out last year, trying to build in too many improvements at once.

Intel hasn’t explained what is causing the defects in the 7nm chips, but it is in the process of adopting a major new manufacturing technology, extreme ultraviolet lithography (EUV.)

EUV tools are enormous, the size of a bus, and cost upwards of $100 million apiece. Delayed for well over a decade by technical challenges, the new EUV tools should enable smaller features on computer chips. But it’s a brand-new process and Intel may be facing unforeseen problems as it works to make EUV work.

Whatever the reason, Susquehanna International Group investment analyst Christopher Rolland said last month it’s time for Intel to throw in the towel.

“We believe it’s near impossible for Intel to catch/surpass (TSMC) in the next half-decade,” Rolland wrote to clients.

“While not probable, we believe it makes sense for Intel to sell some of their leading-edge fabs to (TSMC), given the fungibility of Intel’s state-of-the-art fabs/equipment,” he wrote. “The Taiwanese’s foundry should be able to fix the process issues in relatively short order, and allow Intel to once again focus on CPU architecture, peripheral technologies, and holistic system design advantages.”

Tech insiders have made similar points for years. Intel, though, maintains that integrating chip architecture and manufacturing gives the company an advantage, with its Oregon engineers working inside its Hillsboro research factories to develop new chips and the process to make them in tandem.

Intel asserts that collaboration enables it to move faster, and more cost-effectively, than rivals. And the company has given no indication it’s rethinking that strategy.

Keyvan Esfarjani, the newly appointed head of Intel’s manufacturing group, wrote to his organization Aug. 7 to affirm that integrated design and manufacturing “continues to be a crucial differentiator and advantage for Intel.”

“We get a competitive advantage from the strong connection between products and

process design and will continue to invest in our future process technology roadmap,” Esfarjani wrote. “Also, operating our fabs gives us the ability to control our own destiny in terms of capacity and flexibility for supply.”

“Contingency plans” to outsource some manufacturing is a matter of giving Intel flexibility, Esfarjani wrote. He said Intel’s 10nm production is running 20% ahead of what it expected in January and he insisted the outlook is strong.

“During our company’s history, we have been through intense periods of competition, and this is no different,” he wrote. “Let’s embrace the challenge, and I have no doubt we will win.”

Longtime Intel watcher Jim McGregor, with TIRIAS Research, is among those who has argued in the past that the company split up its architecture and manufacturing. But he said that doesn’t make sense right now, because Intel has too much invested in its current approach.

And McGregor said Intel is still ahead of TSMC and other contract manufacturers in transistor design and materials technology, advances the company can use to get back in the game.

Still, McGregor is mystified that Intel didn’t see the troubles coming with 7nm. He wonders if the company truly didn’t understand things were going wrong, or if the message wasn’t being communicated properly up the chain of command.

“This is unacceptable for this size company, for a technical leader,” McGregor said. “The communication should be clear.”

Intel is on solid footing right now, according to McGregor said, because it is still delivering strong financial results despite its manufacturing setbacks. The company doesn’t haven’t to do anything radical, at least not yet.

Neither Intel’s customers nor its investors will give it too many more chances to get its manufacturing right, though. McGregor said the fundamental question is whether Intel has the right manufacturing strategy.

If it doesn’t, he said, Intel will have to consider some major changes – from rethinking corporate leadership to potentially offloading its factories.

“If it ever does get to the point where it impacts their financial performance,” McGregor said, “then there’s going to be some difficult decisions to be made.”

-- Mike Rogoway | mrogoway@oregonian.com | twitter: @rogoway | 503-294-7699

Subscribe to Oregonian/OregonLive newsletters and podcasts for the latest news and top stories.

If you purchase a product or register for an account through a link on our site, we may receive compensation. By using this site, you consent to our User Agreement and agree that your clicks, interactions, and personal information may be collected, recorded, and/or stored by us and social media and other third-party partners in accordance with our Privacy Policy.