The Great American Microchip Mobilization

That summer Congress passed the CHIPS and Science Act, which earmarks more than $52 billion in grants, loans, and other benefits to chip manufacturers—with the largest portion, $8.5 billion in grants and $11 billion in loans, going to Intel. The bill was born under President Trump and signed by Biden, a vanishingly rare instance of alignment on how to create US jobs and counter China.

The trouble is that Intel has serious baggage. The company was utterly dominant when Windows PCs were the hottest thing in tech, but through a series of blunders, it missed out on building the application processors for what came next: smartphones, then AI. Now, just as the US has started describing Intel as its national champion in the chipmaking arms race, the business ramifications of all those missteps are coming due: Over the past several months Intel’s stock price has plummeted, and it has laid off 15,000 workers, begun a major restructuring, and faced a swirl of rumors that it will be broken up or even sold outright.

Gelsinger says he just needs time to turn the company around. “Intel was a decade-plus in making bad decisions,” he told me. “I always said it was a five-year journey to get us back. We’re at year three and a half.” The question is, will Gelsinger’s plans come together before Intel falls apart? And what about the states and communities that may now be counting on Intel to save them?

For Springowski the news that Intel was coming to Ohio was bittersweet: She’d succeeded in guiding Intel to her state—but not to Lorain. After weighing its options, the company had settled on a huge swath of farmland in New Albany, a small town outside the state capital of Columbus.

Building a microchip fabrication facility, or fab, is a proposition of otherworldly complexity. It requires infra-structure and specialized equipment on a scale that evokes modern-day pyramid building—all for the sake of mass-producing objects measured in nano-meters. And over the next several years, all that monumental stuff would be converging on some fields in central Ohio. The site had a lot going for it, but unlike Lorain, it had no port; the closest one was 140 miles away. So Intel had to figure out how to move everything, mile by slow mile, across half the state.

It’s super early one morning in the summer of 2024, hours before I want to be awake, but also hours before the day will get wicked hot. I’m standing in the very southern part of Ohio near the town of Portsmouth, about 200 miles south of Lorain, on a floodplain above a bend in the Ohio River. I’m in a parking lot across from a gas station, next to a restaurant called Mex-Itali (“The Best of Both Worlds!”). About two dozen guys—and yes, they’re all guys—are milling around me, putting on helmets and yellow safety vests, drinking coffee and smoking. One of them is named Moose.

We’ve all gathered here because about 100 feet away there’s a spot where one two-lane road meets another two-lane road at a 90-degree angle. This would not be interesting on most days, but about an hour from now this enormous, and I mean enormous … thing on 172 wheels will come rumbling up to make the turn.

The guys in the Mex-Itali parking lot are talking about how it’s going to work. Moose runs a quick safety briefing, reminds everyone to hydrate, tells the guys to mind their manners when talking over the radio.

It took nearly two years of planning to get here. This is one of about two dozen “superloads”—pieces of highway cargo that weigh more than 120,000 pounds—that will be lumbering across Ohio for Intel. Today’s load, number 13, is 280 feet long, 23 feet tall, and 20 feet wide. The gargantuan apparatus, weighing nearly 1 million pounds, is called, very anticlimactically, a cold box. It was made by a European company, shipped to New Orleans, brought by barge up the Mississippi and Ohio rivers, and offloaded at a rudimentary -purpose-built port near here—and now it needs to get transported overland to the site that Intel hopes will one day become the largest source of AI chips in the world.

To do that, it turns out you need seven days, a bunch of trucks, a lot of permits, all these dudes—from cable companies, power companies, hauling companies—plus a handful of local cops and highway patrol officers. All along the route, you need to physically move power lines and traffic lights just so the load can make it through. And because it’s so disruptive to the people who live along the route, the biggest superloads need to be delivered before the school year starts.

A cold box, by the way, is part of something called an air separation unit. To make microchips, your factory floor needs to be a cleanroom, because even a microscopic speck of dust can wreck a silicon wafer. So you need to separate air into its component gases and use the nitrogen to remove any other gases, moisture, and particles from all supplies and tools. (Other air components are also used in the manufacturing process.) Eventually, four of these cold boxes will be stood up vertically, like skyscrapers, at Intel’s new, 1,000-acre site.

Along the route, the superloads have a following. People watch for updates from the Ohio Department of Transportation on Facebook. In the comments, many are supportive—one woman offers the crew blackberry cobbler—some are mad about getting stuck in traffic, some are psyched about the logistics of moving something so large. Some want to watch it all go by.

Emily Stone brought her camp chair. Her friends call her the Load Chaser. This is her second superload. “Small town America—this stuff doesn’t happen,” she says. She was born here, grew up here, loves it here. Back in the day Portsmouth had shoe factories, a steel mill, a brickyard. A big plant nearby used to enrich uranium for nuclear weapons—back when that was the technology that superpowers competed over. Stone’s dad worked there for 35 years. He died of leukemia after the plant stopped enriching uranium in 2001. A middle school nearby later closed after testing positive for radioactive material. Stone has been protesting the plant operator’s refusal to take public responsibility.

After most industry died out, Portsmouth became a poster town for the opioid epidemic. At one point it had the highest number of pill mills per capita in the country; it was later the setting of a seminal book about the crisis called Dreamland. In the early 2000s, according to the book, OxyContin served as a local currency. One woman said she was able to buy a car with pills. Stone, who worked as a pharmacy tech at the time, says she knew people who profited from opioids and people who died from opioids.

So for Stone, the superload is a fun distraction, yes. But as it makes its way from one of the poorest parts of Ohio toward one of the richest, she gets why some people in the area are skeptical, or worse. “They’re already on edge,” she says. “And then you have these big, huge loads coming through that nobody really understands.”

“Does anybody even wonder exactly where these superloads came from to begin with?” writes one man on Facebook.

“I’m just disgusted by the whole darn thing,” posts another. “Big business tends to trample on everyone, at every turn, while brainwashing folks into think[ing] they are doing favors for us.”

“They don’t benefit anyone in southern Ohio at all,” writes one more.

Stone and I hear it before we see it—a low but loud hum coming from the west. It’s a giant white box that’s longer than the fuselage of a 747. It takes two semi tractors to move it, one pulling in the front and one pushing from the back. There’s a third just to help push the super-load up hills; it’s on standby the rest of the time. Two guys are posted on a little deck at the back of the load, manning a steering console that helps guide the cold box through tight turns.

I jump in a state car with a spokesman for the Ohio Department of Transportation named Matt Bruning. He’s a former news radio reporter who says AI is now doing advertising voice-overs at one of his old stations. Bruning and Moose have gotten kinda Facebook famous for being part of the crew. “Super Load #13 is making the turn,” Bruning posts. We cruise just ahead of it and bear north into Ohio.

Intel wanted us here. It was their idea that I follow the superload. The company has deployed what feels like a small but happy army of people in matching T-shirts to gin up excitement in the towns along the route, try to deflect annoyed people when traffic backs up, and flood the zone with fun facts. They put up Intel tents, pass out Intel noisemakers and Intel toys to the kids, set up Intel booths next to food trucks. They strap people into VR headsets to show them what working in a fab looks like: engineers in white bunny suits walking around a sterile room full of very expensive machines.

Back with the convoy, we have officially entered the wicked hot part of the day, and we haven’t even made it out of Portsmouth. The superload travels about 6 mph on average, and now a belt has snapped in the engine of the steering console. While a crew works to fix it, the entire convoy has stopped.

People have come out on porches to watch some guys in bucket trucks loosen the big, cantilevered extension arms that suspend stoplights above an intersection. When the load comes through they’ll swing those arms out of the way, then swing them back. One group of spectators is watching from outside a drug rehab center. That’s what Portsmouth is known for these days—a place where people with addiction can come for treatment, counseling, job training, and jobs. The author of Dreamland eventually came back to write a new chapter on Portsmouth, about how the town has remade itself since the worst of the opioid crisis.

Bruning suggests we just stay in the car with the AC on. But then we realize we’re stopped near an old-school joint called Malt Shops that advertises “Ice Cream, Shakes, Sundaes, Sandwiches, and Footers.” A handful of crew guys and Intel flacks and I go in and order some of everything.

The crew is from all over. Joe Jones and his guys drove down from Detroit. He used to work for a company that made parts for Ford. Before Covid, he says, auto work was great. After the chip shortage, “too much stress.” Ford vehicles are “chipped up,” he says, each one with hundreds, even thousands, of microchips. The car can even be deactivated if you haven’t made your payment. Chips, he says, are “what the next war’s gonna be about.” Now Jones works for his cousin’s utility company, helping to raise power lines out of the way of the superload.

Rick and Julia Miller are from northern Florida. He used to supervise roofing crews but got laid off during Covid. She raised high-end chickens that laid -mahogany-brown eggs. A friend told them they could see the country and make money in the pilot-car business. Now the two, who are in their seventies, lead the superload caravan in a truck with orange flags and one of those big yellow signs that says “OVERSIZE LOAD.”

Danny Hoeck is from central Kentucky. He drives the lead semi pulling the superload. He’s been driving trucks for almost 50 years. “I retired two years ago,” he says, laughing. “I just ain’t quit yet.” He says he keeps getting called for jobs because there aren’t enough people qualified to do the work.

And then there’s Moose. He’s big, of course, has a sharp meme game on the socials. His full name is Kieran Drylie, and he’s from New Jersey. We start hearing chatter on the radio that the steering console has been fixed. We finally see the superload inching up the road toward the Malt Shops. Moose’s boss later tells me that rather than taking apart the engine and replacing the broken belt, they just went and got an entirely different engine and installed that instead.

Pat Gelsinger came back to Intel to fix it.

The first time he started working at the company, he was just 18 years old, and Intel was at the very center of the tech industry. Moore’s law—the observation turned prophecy that the number of transistors on a microchip keeps doubling every two years without getting much more expensive—had been coined by his boss, cofounder and CEO Gordon Moore. Intel had invented the world’s very first commercially available microprocessor and helped establish Silicon Valley as a place that wasn’t primarily about fruit trees. People called it one of the “four horsemen” of tech.

Gelsinger was part of all that. He was chief architect of the first processor to contain more than a million transistors. But by the time his first stint at the company ended—when he was “nudged out,” as he has put it, as chief technology officer in 2009—Intel had already started to ride into a ditch.

Intel was still making massive profits in the PC chip market. But in 2006 the company’s then CEO, Paul Otellini, opted to pass up a deal with Steve Jobs to build chips for a new Apple device. Otellini figured its sales would be weak. It was the iPhone. After that, the entire mobile phone sector adopted a chip standard that wasn’t Intel’s.

Then, a couple of years later, a bunch of artificial intelligence researchers started training neural networks—long believed to be a technological dead end—using a chip architecture designed by Nvidia. Their impressive results rang in the breathless current era of AI, along with Nvidia’s dominance of it. (To make matters worse: Over the past 20 years, Intel has reportedly had the chance to acquire major stakes in Nvidia and OpenAI but passed on both of them.)

Like an increasing number of Intel’s competitors, Nvidia was a “fabless” company—one that designed chips and then outsourced their production to chip foundries in Asia, primarily TSMC and Samsung. For the most part, Intel stuck to making its own -leading-edge chips. That might have turned out to be a strong play, says G. Dan Hutcheson, a semiconductor industry expert, but in the early 2010s Intel adopted a new manufacturing technique in its fabs that proved too complex and yielded too few chips.

These errors set Intel so far back that by the end of the 2010s, Hutcheson says, TSMC was the only company in the world that could manufacture the most cutting-edge processors. To the US government, that started to look hugely problematic—because of the company’s proximity to China. Then came Covid and the great chip shortage, which spurred Washington’s sense of urgency even more.

By the time Gelsinger returned as Intel CEO in early 2021, members of Intel’s board were suggesting the company get out of the manufacturing business altogether and just focus on design, says Hutcheson. But Gelsinger was adamant: Intel needed to keep designing and making its own chips on a large scale, while also becoming more of a foundry that manufactures chips for Intel’s “fabless” competitors. In fact, he wanted Intel to become the second-largest foundry in the world by 2030. Gelsinger also set out to remake the company around a new and advanced manufacturing process called 18A, several generations beyond the complicated but low-yield process that got the company in trouble a decade ago.

To do all that, Gelsinger reckoned, he was going to need tens of billions of dollars from the federal government. So he appealed to the rising spirit of economic nationalism in Washington. Four months after he took over as CEO, he wrote an op-ed in Politico arguing that the US had an “urgent need for action to address national security.” He talked about “US chip leadership” and later even called Intel a “patriotic” company. (But only to a point, apparently. Gelsinger’s pitch to the Biden administration was simple: Subsidize US manufacturing or I’ll just build my fabs elsewhere.)

Gelsinger spent much of 2021 in meetings with officials in the US and Europe. In Washington, the CEO lobbied specifically for money to be aimed purely at US companies.

“God decided where the oil reserves are,” Gelsinger likes to say. “We can decide where the fab reserves are.” While Washington dragged its feet, Intel broke ground on two new fabs at its existing site in Arizona, ramped up operations at sites in New Mexico, went looking for new sites around the US, and eventually took Springowski up on her idea to build in Ohio.

The ultimate decision not to build in Lorain was about land. Intel officials said they needed close to 1,000 acres, shovel ready, and Lorain just didn’t have it. So Springowski put Intel in touch with folks who eventually reached out to JobsOhio, a privatized version of the state’s economic development agency. A second agency called One Columbus proposed a swath of farmland in New Albany, a small town outside of the capital that just so happened to be right next to a massive international business park. The deal involved buying out more than 50 landowners. Ohio came in with massive subsidies of its own—about $2 billion for infrastructure improvement, cost offsets, and income tax incentives.