Secretive labs and tightly guarded clean rooms in Hillsboro have long represented the leading edge of semiconductor technology.
That’s where Intel crafted generations of new microprocessors, chips that led the industry for decades as engineers working at atomic dimensions invented new ways of packing more capabilities into a minute space. Those discoveries powered years of progressively faster, cheaper and more advanced computers.
And it’s there, in Hillsboro, that Intel began making these new chips at research factories tethered to its labs. Intel would then send its meticulously developed manufacturing technique to its other factories around the world where it was replicated precisely, a well-established practice called “copy exactly.”
The model led Intel to become Oregon’s largest corporate employer and one of the state’s major economic engines, convening researchers from all over the world to engineer new chips as the company spent billions of dollars on equipment to manufacture their microscopic marvels.
Now, Intel is laying the groundwork to toss the old model out the window. It is openly flirting with the notion of moving leading-edge production from Oregon to Asia and hiring one of its top rivals to make Intel’s most advanced chips.
The company says a decision is likely in January.
It’s a momentous choice that follows a string of manufacturing setbacks at the Ronler Acres campus near Hillsboro Stadium, failures that have cost Intel its cherished leadership in semiconductor technology – perhaps forever.
Outsourcing wouldn’t shutter Intel’s Oregon factories or close down its Hillsboro research labs. The company says it’s committed to maintaining its advanced research and retaining internal production capacity. It’s continuing a massive expansion of its D1X factory in Hillsboro.
In time, though, Oregon’s central role in Intel’s technology would almost surely erode if the company cedes manufacturing leadership to rivals overseas. Chip industry analyst Dan Hutcheson of VLSI Research believes that transition could render Oregon “irrelevant” if Intel gradually shifts away from integrated research and manufacturing.
“Companies say they’re making a transition. What they find is they’re stepping off a cliff,” Hutcheson said. “They’re going down a road that you can’t easily go back on.”
Shocking setbacks
Last summer, Intel shocked investors with the announcement that development of its forthcoming 7-nanometer chip technology is a year behind schedule, now slated to launch in 2023. Intel lost more than $40 billion in market value in one day last July after announcing its 7nm troubles. It has shed billions more in the months since.
Intel suffered similar setbacks with its 14nm and 10nm processes in recent years. The successive stumbles undermined investor and customer confidence that Intel can deliver the regular upgrades in chip technology that made it one of the largest and most profitable companies in the world.
Now, Intel’s own leadership appears to share those doubts. CEO Bob Swan told Wall Street analysts on a conference call earlier this month that it may outsource advanced production to its rivals – he named Taiwan Semiconductor Manufacturing Co., specifically – to ensure “a predictable cadence of leadership products.” Swan told investors to expect a decision by late January.
Intel already outsources as much as a fifth of its production but has kept the leading edge in Oregon. And whatever it decides on outsourcing, Swan said Intel will maintain its advanced research – science performed in Hillsboro – which he described as “a powerful force in creating future differentiation for our products.”
On this month’s analyst call, Swan said Intel believes it can have it both ways – sending advanced production overseas while retaining internal production for components and older products that don’t require the most sophisticated technology. And Swan said Intel believes it could restore advanced manufacturing to its own factories sometime in the future, if it chooses to.
“Many of our future products can no longer be described as manufactured inside or outside or as being a large core or a small core product,” Swan said. “These products will take advantage of hybrid architectural approaches and the universe of (intellectual property) deployed both inside and outside the walls of Intel.”
Others are dubious.
“No one’s been able to make it work that way. That’s like having two spouses,” said Hutcheson. He said it’s too expensive to maintain internal factories while outsourcing the most valuable technology.
“Once they start going down this path it’s almost impossible to go back because it creates its own momentum,” Hutcheson said. “For me, it’s just a real, real serious decision for (Intel) to make.”
If Intel proceeds, that would serve as a clear indication that the U.S. has lost the leading edge in semiconductor manufacturing. And its decision comes at a tricky moment, geopolitically. The Trump administration has been assertive about bolstering the domestic semiconductor industry, blocking foreign investment in U.S. chip companies and sanctioning Chinese chipmaker Huawei.
TSMC said last spring that it will build a factory in Arizona, bolstered by unspecified state and federal incentives, but the details of its plans appear up in the air and may depend to some degree on the outcome of next month’s presidential election. Any new Arizona fab is many years away, so new chips TSMC makes for Intel would come from Taiwan.
A painful era
Intel is Oregon’s largest corporate employer, with 21,000 employees at its campuses in Washington County. The company’s headquarters are in Silicon Valley but its largest and most advanced operations are in Oregon.
In addition to its research labs and factories – which the chip industry calls fabs – Intel maintains corporate offices in Oregon that employ thousands of administrative and marketing personnel. And the company spends billions of dollars to build and equip its fabs, creating a huge ecosystem of suppliers, contractors and construction workers who build and maintain its fabs.
Hillsboro and Washington County have granted Intel property tax exemptions that saved the company more than $1.1 billion over the past decade. The tax breaks are some of the nation’s biggest but they’re tied to the value of Intel’s capital spending in Oregon, so the company only saves money on its taxes if it continues to invest.
There’s no sign the company is slowing down. The company had a shortage of manufacturing capacity for much of 2019, just opened a huge leading-edge factory in Arizona, and is nearly two years into a multibillion-dollar expansion of its enormous D1X research factory in Hillsboro.
“Our four Oregon campuses are Intel’s largest concentration of talent and facilities in the world,” Courtney Martin, Intel public affairs director, said in a written statement this past week. “We continue to grow and invest in our Oregon facilities, which are at the heart of Intel’s research and development and our future.”
Indeed, Intel’s Oregon work will continue uninterrupted whatever Intel decides in January. But it’s the advanced research and manufacturing that serves as the center of gravity for Intel’s Oregon operations, and if Intel sends the production work elsewhere it will inevitably diminish the state’s role within the company.
“It’s something that throws the future of the fab operations in Oregon up in the air a bit. It’s not like TSMC could take over all of Intel’s fab business, but it could take a chunk of it,” said Kevin Krewell, principal analyst with Tirias Research.
If Intel does outsource advanced manufacturing, Krewell said, it could begin a painful era for the entire company as it works to straddle internal manufacturing and outsourcing. Ultimately, though, Krewell said Intel’s advanced research is likely to stay in Oregon – home to the company’s top researchers and most advanced factories.
“That’d be hard to move. I think when you’ve got a dedicated research fab, and you’ve got the team there working on developing the advanced nodes, I think it’s hard to migrate them to another location,” Krewell said.
Electronics manufacturing has been migrating to Asia for decades, lured initially by low labor costs overseas. Labor, though, makes up a small portion of semiconductor manufacturing costs. Most of the expense is in manufacturing tools – it costs billions of dollars to build and equip a fab, and those costs are roughly the same whether the facility is in Hillsboro, Oregon or Hsinchu, Taiwan.
It’s not cost, then, that’s prompting Intel to look for production overseas. It’s quality.
Intel has had a recurring problem with the “yields” on its advanced processors, meaning that too many chips coming off its production lines have defects that render them useless.
If a small share of chips is defective, Intel can just toss them out and write off the loss. With low yields, though, the volume of defects is so high that Intel cannot make money off those that remain.
When Intel ran into trouble with its current generation of 10nm chips, the company said it was too ambitious with advances and it assured investors and customers that it would take a more measured approach with the forthcoming 7nm chips.
And yet the 7nm process ran into trouble, too. On some level, Intel’s issues simply reflect the fact that advances in chip technology become extraordinarily difficult as the features on the chip become infinitesimally small.
What’s striking, though, is that TSMC hasn’t had similar problems meeting its target dates for new chips.
“I assume Intel has done a lot of internal soul searching on this. Each process node is a challenge,” Krewell said. “Maybe TSMC has been a bit luckier by taking a more conservative route. They don’t get as much benefit, but they do get enough benefit.”
Weston Twigg, a Portland investment analyst who follows Intel for KeyBanc Capital Markets, said he sees opportunities for Intel if it chooses to segment its manufacturing. The company could send advanced microprocessor production to TSMC while making less advanced complementary technology that could be packaged with the processor.
And while Intel faces a tough choice, Twigg said it comes at an opportune time. The semiconductor market is expanding, with new categories emerging in artificial intelligence, embedded devices and other emerging technologies. Those other segments don’t necessarily require the most advanced manufacturing, Twigg said, so Intel can make some products in its existing factories while sending cutting-edge products to Taiwan.
“It would be hard to manage that kind of relationship where you’re bouncing back and forth,” Twigg said. “But I think they do want to have, more broadly speaking, more manufacturing flexibility.”
Ultimately, Twigg said Intel has put itself in an unenviable position. The company has lost the competitive advantage in production technology that set it apart for years. Now, he said Intel is seeking a new strategy that makes the most of what is left.
“Intel’s still searching for what it looks like,” Twigg said. “But it’s not the Intel from 2005 model, which doesn’t work anymore.”
-- Mike Rogoway | mrogoway@oregonian.com | twitter: @rogoway | 503-294-7699
The Link LonkNovember 01, 2020 at 09:25PM
https://ift.tt/3mJL7k9
Intel contemplates outsourcing advanced production, upending Oregon’s central role - oregonlive.com
https://ift.tt/2YXg8Ic
Intel
No comments:
Post a Comment