Now that Pat Gelsinger is no longer president of the troubled chipmaker Intel, he can acknowledge one fact of the semiconductor industry: Nvidia is far ahead of its competitors in technology.

"Look, they've executed well… and ultimately, [Nvidia CEO Jensen Huang] is pushing his team to stay ahead," the former Intel CEO said on Yahoo Finance's Opening Bid podcast.

"They've built a meaningful moat around their franchise," he added.

For more than three years, Gelsinger has aggressively led Intel's turnaround. He laid off thousands of employees, cut costs, secured funding from the CHIPS Act, built chip foundries, and promised fast AI chips that can compete with Nvidia and AMD (eg. AMD Parts XC7A200T-2FFG1156I).

He was fired in early December after missing financial targets, lack of progress on AI chips, and cash hemorrhaging from the foundry business.

Intel's fourth-quarter sales fell 7% year-over-year to $14.3 billion, and net income plunged 76%. The company expects to only break even this year.

Analysts believe Nvidia could generate more than $5 in earnings per share by 2025. Nvidia's stock price has risen 1,220% over the past five years, while Intel's stock price has fallen 69%.

Intel announced Lip-Bu Tan as its new CEO in March.

Tan was CEO of Cadence Design Systems (CDNS) from 2009 to 2021. After serving on Intel's board for two years, he left in August 2024 due to a conflict with Gelsinger over the positioning of the business.

Tan often pushed for a better artificial intelligence strategy to counter Nvidia and speed up decision-making at Intel, which is known for its bureaucracy.

Gelsinger said no one could replace Nvidia.

"I've always believed there's a 'rule of 10,' where if you're not 10 times better than the best, you can't replace him. Now, you have to be at least 10 times better consistently so that people say, well, yes, I'm going to invest in it," Kissinger added.

Chen will publicly outline his strategy and detail how he will respond to Trump's tariffs when the company reports first-quarter results after the market closes on April 24.

Pat Gelsinger once said that Jensen was lucky in AI

At GTC 2025, former Intel CEO Pat Gelsinger reiterated his oft-repeated claim that Nvidia CEO Jensen Huang "got lucky" in the AI revolution, but he went a little deeper to explain his rationale.

As GPUs became the focal point for AI innovation, Nvidia is now one of the world's most valuable companies, while Intel has struggled. But it wasn't always this way. 15 to 20 years ago, Intel CPUs dominated computing because they were able to handle all major workloads. During this time, Intel missed the boat on AI and high-performance computing (HPC) because Project Larrabee attempted to build a GPU using the x86 CPU ISA. In contrast, Nvidia is betting on purebred GPUs, said Pat Gelsinger, former Intel CTO and CEO, at Nvidia's GTC 2025 conference.

"[In the mid-2000s] the CPU was king, and I admired Jensen for insisting, 'No, I'm not trying to build that CPU; I'm just trying to solve workloads starting with graphics processing,' " Gelsinger said. "You know, it became a broader vision. Then he got lucky with AI, and I was arguing with him one time, and he said, 'No, I got lucky with AI workloads because it requires that kind of architecture.' That's where the center of application development is [now]."

One of the reasons Larrabee was canceled as a GPU project in 2009 was that it was not competitive as a graphics processor compared to graphics solutions from AMD and Nvidia at the time. In part, this was because Intel wanted Larrabee to be extremely programmable, which resulted in it lacking key fixed-function GPU parts such as raster operations units. This affected performance and increased the complexity of software development.

"I had a well-known project called Larrabee that tried to combine the programmability of the CPU with the throughput-oriented architecture of the GPU. I think if Intel had stayed on that path, the future might have been different," Gelsinger said during the webcast. "I give Jensen a lot of credit for sticking with the vision of throughput computing or acceleration."

Unlike AMD and Nvidia GPUs, which use proprietary instruction set architectures (ISAs), Intel's Larrabee uses the x86 ISA with Larrabee-specific extensions. This provides an advantage for parallel general-purpose computing workloads, but is a disadvantage for graphics applications. As a result, Larrabee was relaunched in 2010 as the Xeon Phi processor, originally designed to handle supercomputing workloads. However, Larrabee did not gain much traction as traditional GPU architectures gained general-purpose computing capabilities through the CUDA framework and the OpenCL/Vulkan and DirectCompute APIs, which are more easily scalable in terms of performance. Intel abandoned the Xeon Phi project between 2018 and 2019 after the Xeon Phi "Knights Mill" failed to meet expectations, moving instead to data center GPUs for high-performance computing (HPC) and specialized ASICs for artificial intelligence (AI).

Larrabee and its Xeon Phi family successors failed in large part because the CPU ISA they were based on didn't scale well for graphics, AI, or HPC performance. Larrabee's failure began in the mid-2000s, when CPUs still dominated and Intel's technology executives saw x86 as the way to go. Today, Intel's attempts to adopt more traditional GPU designs for AI have largely failed, and the company recently canceled its Falcon Shores GPU for data centers. Instead, the company is pinning its hopes on the next-generation Jaguar Shores, which isn't expected to be released until next year.

Source: Content compiled from Yahoo

View more at EASELINK