How innovation died at Intel as it faces an uncertain future as America’s only leading-edge chip manufacturer

As the rise of Arm-based chips and AI GPUs eroded Intel’s standing, Pat Gelsinger promised to turn things around by launching a foundry business — leveraging Intel’s manufacturing capabilities to make chips for external customers.

The move was intended to bring back bold, innovative thinking — a so-called “Grovian” mindset, referring to its famed former CEO Andy Grove.

Some say it was Gelsinger’s best bet.

“I think it’s the right move because without looking outside the company and pursuing external customers, it’s a shrinking company,” Hari said.

Others were skeptical of the approach. Multiple analysts and insiders said the decline of Intel’s product business handcuffed its ability to fund Gelsinger’s aggressive capital spending and hiring sprees — which gave way to layoffs that depressed morale.

Several employees commended what they described as Gelsinger’s brilliance and good character — one posted on LinkedIn about Gelsinger helping his grandson access medications to treat a fatal genetic disease.

Others contended that Gelsinger set expectations too high, then acted more as a cheerleader than executor and refused to listen to input that didn’t align with his vision.

Gelsinger’s stated target when he became CEO was to have Intel’s foundry generating $15 billion in revenue by 2030 — roughly double the $8.5 billion Samsung generated last year after nearly two decades of operation, according to Gartner data cited by Rasgon. In its most recent quarter, Intel’s net loss totaled $16.6 billion due to substantial losses in its chip manufacturing division.

“Intel has not demonstrated that they can execute on foundry … because they don’t have a track record and the track record that you’ve seen from their internal execution has been so flawed,” KeyBanc analyst John Vinh told Yahoo Finance. “It’s hard for anyone to commit any major mission-critical applications for them. And as a result of that, Intel is not going to pick up any sort of meaningful business.”

Its last hope to right a sinking ship has been called into question. Under Gelsinger, Intel staked its fate on a new advanced manufacturing process called 18A — which the company’s corporate vice president Bruce Andrews told the Financial Times in November would “bring (the company) back to technological leadership.” The process would allow Intel to manufacture the most advanced AI chips with tiny components tens of thousands of times smaller than the size of a human body cell.

Intel initially said it would begin high-volume production with its 18A process in the first half of 2025 — then moved that target to the second half of next year, according to its third quarter earnings call.

Internal testing of 18A earlier this year reportedly showed it wasn’t ready for high-volume production.

A current employee at one of Intel’s fabs told Yahoo Finance there are “a lot of issues” making chips with the 18A process. They said that Intel is not ready to take on external customers — and that communications between teams that should take days often take weeks as employees avoid responsibility for mishaps.

Intel declined a request to comment on 18A.

Meanwhile, Intel has already purchased at least $760 billion worth of machinery for its successor manufacturing process to 18A, called 14A.

Gelsinger was ousted on Dec. 1 by Intel’s board, which has been criticized as “extremely weak” and lacking “hardened semiconductor people” by sources Yahoo Finance spoke to. Notably, the board added two leading semiconductor experts days after Gelsinger’s exit.

Intel’s technology is, in large part, responsible for the digital revolution. Its co-founder Bob Noyce has the greatest claim to the founding of Silicon Valley, according to semiconductor expert Chris Miller, author of “Chip War.” The company invented the world’s first microprocessors (i.e. computer chips) and the x86 architecture (a critical blueprint for designing computer chips).

Its other co-founder, Gordon Moore, created “Moore’s Law,” a theory that defined the pace of innovation in the semiconductor industry for more than half a century.

But after the dot-com bust, Intel invested in multiple projects that never materialized or failed to reach their potential. Two former executives told Yahoo Finance that innovative efforts were often killed if they didn’t immediately contribute to revenue or risked cannibalizing existing products.

One former high-level executive, who worked within several divisions, said Intel didn’t support the team working on low-power Atom chips for mobile phones in the early 2000s. It sold its license for Xscale, then Arm’s most advanced architecture for mobile chips, to Marvell (MRVL) in 2006.

Paul Otellini, who served as CEO from 2005 to 2013, also passed on making chips for the initial Apple iPhone. Instead, Intel bet on Nokia — “a spectacular failure in terms of strategic decision making,” a former executive said.

Its manufacturing division opted to use a technique called multiple patterning rather than investing in EUV (extreme ultraviolet) lithography machines — which use incredibly complex technology from Dutch company ASML (ASML) that was only made possible by funding from Intel in the late 1990s and early 2000s.

TSMC and Samsung (005930.KS) successfully adopted EUV lithography first, vital to today’s advanced chipmaking.

Six former executives pointed to poor leadership as the source of many problems. They said Gelsinger’s predecessors Brian Krzanich (known as “BK”) and Bob Swan prioritized short-term thinking over long-term technology strategies.

The two served from 2013 to 2018 and 2018 to 2021, respectively.

Intel fumbled multiple attempts to enter what would become the AI chip market. In 2009, it scrapped a project codenamed Larrabee, led by then-chief technology officer Gelsinger, which aimed to develop a standalone GPU like Nvidia’s.

In 2017, Intel hired AMD’s graphics chip engineer, Raja Koduri, to lead a second effort toward a homegrown GPU. Three former executives say Koduri had a strong vision but was weak on execution.

Koduri told Yahoo Finance, “As member of executive leadership team, I did bear the accountability of overall execution. Being criticized given my responsibility is true but, given the factors at play, is unkind to the engineering team that executed against all odds.”

At the same time, Intel acquired Habana Labs to develop another type of AI chip called an ASIC accelerator. But the effort wasn’t prioritized due to internal politics, according to an executive who helped oversee the acquisition.

“Their focus on Nvidia, who is the real enemy, was simply not there,” the source said. “They were like …crabs fighting amongst themselves.”

To add insult to injury, Intel missed opportunities to acquire Nvidia and invest in OpenAI.

Meanwhile, rivals were improving rapidly. TSMC began manufacturing iPhone chips in 2014, and Apple ditched Intel after creating its own MacBook chips using Arm’s architecture. Nvidia spent nearly two decades developing the technology used in its AI chips before they took off in earnest. ASML spent 17 years creating its EUV lithography systems that gave it an effective monopoly over the tech required to make Nvidia chips.

By the time generative AI sparked a new era of tech boom, Intel had been left in the dust.

Chipmaking is becoming an increasingly important national security issue as US-China relations deteriorate. The US government is betting on Intel, sinking nearly $8 billion in CHIPS Act funding into the company’s existing and future foundries.

But the money adds to Intel’s conundrum: Its product business would be better off outsourcing all chipmaking to TSMC, Hari said, but that would leave its manufacturing division with no revenue. Analysts from Citi, KeyBanc, Raymond James, and others say Intel should spin off or divest its foundry business, but it has to maintain a 50.1% stake to keep its CHIPS Act funding.

“There don’t seem to be any easy answers here, so whoever winds up filling the (CEO) slot looks in for a tough ride,” Rasgon wrote in a client note.

Intel’s revenue for the upcoming quarter is expected to fall more than 10% year over year to $13.8 billion, per Bloomberg consensus estimates, and earnings per share are projected to drop 77% to $0.12.

The company named CFO David Zinsner and former head of client computing Michelle Johnston Holthaus as interim co-CEOs, with Holthaus heading the newly formed Intel Products division. Two former high-level executives critiqued the move, with one saying Holthaus has “no technical skill.”

” To put these co-CEOs in shows you they had no succession plan,” Rasgon told Yahoo Finance. He said the CEO ” everybody wants” to see is Lip-Bu Tan, the former CEO of Cadence Design Systems, who is reportedly a candidate. Tan also sits on the board of Gelsinger’s Christian organization.

Zinsner and Holthaus did not respond to Yahoo Finance’s requests for comment.

During UBS’s Technology and AI Conference in Arizona on Dec. 4, Zinsner said Intel remains committed to its foundry business. ” (We) still want to be a world-class foundry, want to be the western provider of leading-edge silicon to customers,” he said.

One former executive said the best pick for Intel’s permanent CEO should be a ” hardcore” chips expert who understands the technology. Another agreed — but indicated the person needs to be an outsider.

“Nobody in the Intel Technology Development Group, who’s either at the top level or one level down, would even have a seat at the table anywhere in the first three levels of management at TSMC,” said the former executive who worked in Intel’s foundry.

While the government is funding the expansion of TSMC’s and Samsung’s foundries in the US, the vast majority of those companies’ manufacturing capacity is in their home countries of Taiwan and South Korea, per Miller. That’s seen by some as a national security risk in case of a Chinese invasion of Taiwan. But “that doesn’t automatically make Intel the solution,” Rasgon said.

” We can throw all the money we want at Intel, it doesn’t necessarily fix the problem,” he added. Nearly 80% of analysts tracked by Yahoo Finance, including Rasgon, have a Hold rating on the stock, with an average price target of $24 over the next 12 months.

“I don’t envy the new CEO,” one former executive said. “How many leading-edge silicon providers can the world afford? Is it two or is it three? If it’s three, then you’d say, ‘OK, as long as they get their manufacturing technology up, it’ll be OK.’ If it’s two, then … who’s going to die as a foundry, Intel or Samsung?”

Yasmin Khorram is a senior reporter at Yahoo Finance. Follow Yasmin on Twitter/X @YasminKhorram and on LinkedIn. Send newsworthy tips to Yasmin: yasmin.khorram@yahooinc.com

Laura Bratton is a reporter for Yahoo Finance. Follow her on X @LauraBratton5

Click here for the latest technology news that will impact the stock market

Read the latest financial and business news from Yahoo Finance