Intel CEO Recalls ‘Bankrupt’ Fears and the Fight to Win Back Top Talent
Table of Contents
The Struggle to Hire Amidst Crisis
For Lip-Bu Tan, the early days of his tenure as Intel CEO weren’t just about fixing a broken manufacturing pipeline—they were about convincing people that the company actually had a future. Speaking at the JP Morgan Global Technology, Media and Communications Conference on Tuesday, Tan candidly described a period where Intel’s financial instability had become a deterrent for the industry’s most sought-after engineers and executives.
According to Tan, potential recruits were blunt about their hesitations. “I tried to recruit some talent,” Tan noted. “They said, ‘It’s almost a bankrupt company, why should I join you?'”
The perception of fragility was a direct reflection of a balance sheet that had been battered by years of missed deadlines and lost market share. To counter this, Tan focused on stabilizing the company’s financial foundation. This included securing equity investments via the Trump administration’s CHIPS program and leveraging high-level industry ties. Notably, Nvidia CEO Jensen Huang committed $5 billion and Softbank’s Masayoshi Son, a former Intel board member, stepped in as a backer.
Tan suggests these moves did more than just inject cash; they provided a signal of confidence to the talent market. The improved liquidity eventually allowed Intel to buy back a stake it had previously sold to Apollo, a move designed to curb earnings-per-share dilution.
Beyond 14A: The Race to Sub-Nanometer
With the financial house somewhat in order, Tan is now pivoting toward a long-term technical offensive. While the industry has been focused on the 18A process, Tan is already signaling a roadmap that extends into 10A and 7A nodes. In the world of semiconductor fabrication, these angstrom-level designations represent a push toward sub-nanometer technology, as 10 angstroms equal 1 nanometer.
Tan believes that a single breakthrough node is not enough to lure back major fabless customers who are currently beholden to TSMC. “People don’t go to you just for one node. They’re looking for the roadmap for the future,” he explained, emphasizing a strategy built on operating efficiency and defect density.
The current status of Intel’s immediate goals appears optimistic. Tan reported that the 18A process is seeing a 7 percent monthly yield improvement, while the subsequent 14A node is reportedly ahead of its year-end target. However, the timeline for the more advanced 10A and 7A nodes remains distant. Tan indicated that risk production for 14A is slated for 2028, with volume production in 2029—paralleling TSMC’s A14 timeline. This suggests that the 10A and 7A technologies likely won’t hit the market until after 2030.
The Agentic AI Pivot
Central to Intel’s recovery is a bet on the evolving nature of AI. For the last several years, the AI boom has been almost exclusively a win for GPU manufacturers like Nvidia, as training massive models requires immense parallel processing power. However, Tan argues that the tide is shifting toward “agentic AI” and inference workloads.
In these scenarios, where AI agents perform autonomous tasks, the demand for CPUs increases. Tan claims that the traditional ratio of one CPU to eight GPUs for training is shifting. “Startups all tell me, Lip-Bu, CPU actually is more useful, even single-threaded,” Tan said, suggesting that some workloads are moving toward a 1:1 ratio or even 4 CPUs to 1 GPU for inference.
Rebuilding the Foundry Culture
Intel’s transition from a company that only makes its own chips to a foundry service for others is a massive cultural and operational shift. To bridge this gap, Tan has leaned into external expertise, poaching Shawn Han, a 30-year veteran of Samsung Foundry, as SVP of Foundry Services.
The demand for these services seems to be materializing in unexpected ways. Tan revealed that some customers are so eager to secure capacity and packaging technology—specifically Intel’s EMIB-T—that they are offering to prepay for wafer substrate materials. “They jump on it,” Tan said, claiming these commitments involve billions of dollars over the coming years.
Despite the optimism, the numbers show a company still in the midst of a painful transition. Intel reported a loss of $267 million on revenues of $52.9 billion during 2025, a slight improvement over the $18.8 billion loss from the previous year, but a reminder that the road to profitability remains steep.
