Intel's new CEO plots overhaul of manufacturing and AI operations

Summary:

• Tan to focus on Intel Foundry and AI chip production, sources say
• Tan aims to improve efficiency and revive Intel's manufacturing prowess, sources say
• New Intel CEO plan likely to include management reorganization, sources say
• Tan is considering broadening Intel's AI business, sources say 

Intel's incoming CEO Lip-Bu Tan has considered significant changes to its chip manufacturing methods and artificial intelligence strategies ahead of his return to the company on Tuesday, two people familiar with Tan's thinking told Reuters, in a sweeping bid to revive the ailing technology giant.

The new trajectory includes restructuring the company's approach to AI and staff cuts to address what Tan views as a slow-moving and bloated middle management layer. Revamping the company's manufacturing operations, which at one time only made chips for Intel but have been repurposed to make semiconductors for outside clients such as Nvidia, is one of Tan's core priorities, these sources said.

The plans are still being formulated and could yet change, these sources said.

Intel shares rose more than 8% in mid-day trading on Nasdaq.

At a town hall meeting following his appointment as CEO last week, he told employees that the company will need to make "tough decisions," according to two other people briefed on the meeting.

Semiconductor industry expert Dylan Patel said a big problem under former Intel CEO Pat Gelsinger, who left the company in December, was that he was "too nice." "He did not want to fire a bunch of middle management in the way they needed to," he said.

Tan, 65, former CEO of chip design software firm Cadence and tech investor, was a member of Intel's board until he resigned last August. In returning as CEO, Tan is set to take over the American icon after a decade of bad decisions by three CEOs in which it failed to build chips for smartphones and missed surging demand for AI processors, allowing competitors Arm Holdings and Nvidia to dominate those markets.

Intel reported an annual loss of $19 billion in 2024, its first since 1986.

In the near term, Tan aims to improve performance at its manufacturing arm, Intel Foundry, which makes chips for other design companies such as Microsoft and Amazon, by aggressively wooing new customers, according to the people.

It will also restart plans to produce chips that power AI servers and look to areas beyond servers in several areas such as software, robotics and AI foundation models.

"Lip-Bu will be spending a lot of time listening to customers, partners and employees as he comes on board and works closely with our leadership team to position the business for future success," an Intel spokesman said in a prepared statement.

Intel declined to comment further or make Tan available for an interview. Tan's venture firm, Walden Catalyst, did not respond to requests for comment.

At the outset, Tan's strategy appears to be a fine-tuning of that of Gelsinger. The centerpiece of Gelsinger's turnaround plan was to transform Intel into a contract chip manufacturer that would compete with Taiwan Semiconductor Manufacturing Co., or TSMC, which counts Apple, Nvidia and Qualcomm as customers.

Gelsinger committed tens of billions of dollars to build factories in the US and Europe to make chips for both Intel and for outside customers, but he was forced to scale back those ambitions as the market for Intel's core products cooled.

BETTING ON AI

Tan has been a vocal internal critic of Gelsinger's execution, according to the two sources familiar with Tan's plans.

For most of its history, Intel has manufactured chips for only one client - itself. When Gelsinger became CEO in 2021, he prioritized manufacturing chips for others but fell short of providing the level of customer and technical service as rival TSMC, leading to delays and failed tests, former executives have told Reuters.

Tan's views were shaped by months of reviewing Intel's manufacturing process after the board in late 2023 appointed him to a special role overseeing it, according to a regulatory filing.

In his assessment, he expressed frustration with the company's culture, sources told Reuters, saying it had lost the "only the paranoid survive" ethos enshrined by former CEO Andy Grove. He also came to believe that decision-making was slowed down by a bloated workforce, Reuters reported.

Tan presented some of his ideas to Intel's board last year, but they declined to put them into place, according to two people familiar with the matter. By August, Tan abruptly resigned over differences with the board, Reuters reported.

When he returns as CEO this week, he will lay fresh eyes on Intel's workforce, which was slashed by roughly 15,000 to almost 109,000 at the end of last year, the sources said.

Beyond the cuts, Tan has little choice but to make Intel's existing manufacturing operation work in the short run. Intel's next generation of advanced chips equipped with AI features, called Panther Lake, will depend on its in-house factories using a new set of techniques and technologies Intel calls "18A."

Intel's financial success this year is tied to strong sales of the forthcoming chip.

Tan signalled in a memo Intel published Wednesday that he plans to keep control over the factories, which remain financially and operationally separate from the design business and restore Intel's position as a "world-class foundry."

Intel's contract manufacturing operation can succeed if Tan wins over at least two large customers to produce a high volume of chips, industry analysts and Intel executives told Reuters.

Part of the effort to lure large customers will involve improving Intel's chip manufacturing process to make it easier for potential customers like Nvidia and Alphabet's Google to use.

Intel has demonstrated improvements in its manufacturing processes in recent weeks and has attracted interest from Nvidia and Broadcom that have launched early test runs, Reuters reported. Advanced Micro Devices is also evaluating Intel's process.

Tan is expected to work on ways to improve output or "yield" to deliver higher numbers of chips printed on each silicon wafer as they move to volume manufacturing of its first in-house chip using the so-called 18A process this year.

The goal is to move to an annual release schedule of AI chips, similar to Nvidia, but that will take years. It will be at least 2027 before Intel can develop a compelling new architecture for a first AI chip, according to three industry sources, and one person familiar with Intel's progress.