Former CEO Pat Gelsinger reveals he was 'not given the opportunity' to finish his job at Intel as he predicts the future of computing will be a 'trinity' of traditional, AI and quantum machines

“The decision to step down from Intel was an extremely difficult one. I wanted to finish what I started, but as you know, I was not given the opportunity,” says ex Intel CEO Pat Gelsinger in a new interview with Japanese outlet PC Watch. This is the first hint from Gelsinger we’ve seen of what many have long suspected, specifically that his departure from Intel may not have been the simple “retirement” that was claimed at the time.

Gelsinger was attending a press conference in Japan held by Playground Global, a Silicon Valley investment fund. The ex-Intel CEO now directs the fund’s investment into semiconductor businesses, which loosely translates into companies that make computer chips.

Along with the comments he made about his departure from Intel, Gelsinger also described his vision of the future of computing. “I believe that conventional, AI, and quantum will continue to coexist as a trinity,” he said.

“Conventional computers have algorithms that are suited to them, AI has algorithms that process high throughput, and I believe that quantum computing will be introduced as a new algorithm that cannot be calculated by conventional computers or AI.”

Gelsinger also explained that AI inferencing as opposed to training may not necessarily be best suited to GPUs, such as those made by Nvidia. In other words, while GPUs are the best option for training AI models, they may not be the optimal hardware for delivering AI services.

A photograph of Intel's Interim Co-CEO Michelle Johnston Holthaus standing on stage, with a background displaying Panther Lake and Intel 18A

The success of Panther Lake and 18A, due later this year, will arguably dictate how history views Gelsinger’s tenure at Intel. (Image credit: Future)

“AI is still in the early stages of growth, and I think inference in particular is a different market from learning. And it’s inefficient to use a solution for learning for inference,” Gelsinger said.

Presumably, Gelsinger is helping Playground Global identify what that alternative to GPUs will be as an investment opportunity. Among other investments in this area, Playground Global has a stake in Snowcap Compute, which is developing computer switches with super conductivity properties as an alternative to traditional CMOS semiconductors.

Snowcap Compute claims such switches can operate 10 times faster than conventional transistors while offering 25 times the power efficiency, despite the extreme helium cooling required for its chips to operate. Snowcap Compute already has chips running at 30 GHz. Playground Global is also invested in PsiQuantum, which has plans to create a one million QuBit (quantum bit) quantum computer by 2027.

All that should keep Gelsinger busy. In the meantime, we’ll have to wait and see how history views his tenure at Intel. Perhaps the most significant milestone will be the release of Intel’s Panther Lake CPUs on the all important 18A node.

Back in the day, Gelsinger himself said he’d bet the company on 18A. If it turns out to be a great success, Gelsinger will look like a visionary hero who was pushed out of the company to soon. If 18A flops, well, best not to think about that for now. It won’t be pretty, for Intel or Gelsinger’s legacy.