In July 2026, former Intel CEO Pat Gelsinger appeared on the All-In Podcast for a nearly hour-long, candid conversation with host Jason. He laid bare the critical missteps that led Intel from its PC-era dominance to being left far behind by Nvidia and TSMC.
From the pinnacle of the PC chip market to being vastly outpaced by competitors, what were the decisive wrong turns Intel took over the past decade-plus? Gelsinger, drawing on 34 years of personal experience, directly identified the root causes of Intel's "derailment": putting financiers in charge, replacing factory investments with hundreds of billions in buybacks, and missing the iPhone and AI chip waves. This reflection exposes not just a giant's technological miscalculations, but a deeper issue the entire tech industry is grappling with: the immense cost when capital returns dictate technological decisions.
Gelsinger recalled that the Intel he joined as a young man was led by a group of "professional technologists"—Andy Grove, Gordon Moore, and Bob Noyce—who were not just business leaders, but engineers among engineers. When he first entered the senior management team, "15 out of 20 people were PhDs."
This technical DNA gave Intel the courage to make long-term, capital-intensive investments. Decisions like building a chip factory or purchasing EUV lithography machines might not show a return on the balance sheet for a decade, but technically minded leaders could see their strategic value. It was this foresight that established Intel's "Intel Inside" dominance in the PC era.
When asked "what went wrong," Gelsinger offered a blunt list. He stated that Intel's "derailment" began when the company started being run by "business people" rather than "technologists." When he returned in 2021, he was "the first CEO with a technical background in 15 years." Previous CEOs either came from the business side or had a finance background.
"When you're making core technology decisions that affect billions of dollars, you can't do it through a spreadsheet. Unless the technology trends themselves make the investment justifiable, it's doomed to be a bad deal." He pointed out that great tech companies like Microsoft's Nadella and Google's Pichai, even if not founders, are "deeply technically literate leaders." Under non-technical leaders, decision-making at Intel became "number crunching," not "looking at the future."
Gelsinger revealed that in the five to six years before his return, Intel returned approximately $100 billion to shareholders through dividends and stock buybacks. When he took over in 2021, he found himself leading a chip company that hadn't built a new wafer fab in a full decade—and hadn't even purchased a single EUV lithography machine. "If that money had stayed on the balance sheet, we could have done many things—like making chips for the iPhone (Intel had turned down Apple) or getting a head start in the foundry business." He sharply criticized this as Intel's "financial short-sightedness" pain point.
During Intel's CPU glory days, the company "looked down on Jensen Huang's graphics cards," dismissing them as "toys for gamers." But Nvidia continuously built its CUDA software stack and SIMT programming model, transforming the GPU into a general-purpose computing platform. The turning point came when Japanese HPC teams started using graphics cards for high-performance computing.
Gelsinger noted that Intel internally had a similar project, Larrabee, aimed at using the x86 architecture for the same kind of general-purpose computing. "But a week after I first left Intel, that project was canceled. If we had persisted, the world would be a very different place."
In 2006, Intel declined a request to develop the chip for the original iPhone, citing unfavorable price and volume projections. Apple turned to ARM instead, launching the mobile computing era while Intel was completely left behind. Even more telling, Steve Jobs had already secretly ported macOS to the x86 architecture "for four versions" while collaborating with Intel on Mac chips. While Intel was proud of winning the Mac business, Jobs was preparing to "ditch Intel." Gelsinger recalled Jobs saying, "I've ported it for x86 for four versions. I think we can handle it." The seed for Apple's own chips was planted there, ultimately leading to the M-series and Intel's complete removal from Apple's lineup.
Intel long adhered to the IDM (Integrated Device Manufacturing) model, making chips only for itself and never considering opening its factories to third parties. TSMC, from the start, positioned itself as a "pure-play foundry," willing to manufacture anyone's designs. When Gelsinger returned in 2021, TSMC's wafer output was already five times that of Intel. "The industry had changed—everyone wants massive wafers, diverse designs, and standardized EDA tools. And Intel was still building in a closed system."
After returning as CEO in 2021, Gelsinger launched the "IDM 2.0" strategy, with the core being to make Intel a foundry and open its manufacturing capacity to external customers. He also actively promoted the CHIPS Act to secure US government funding for building new fabs domestically. He noted that the US's global share of advanced process manufacturing has risen from 12% in 2021 to 18% now, "still a long way from 50%, but the direction is right." Intel's own foundry business is also starting, and TSMC's US factory has begun production.
Regarding the current AI infrastructure boom, Gelsinger believes the bubble won't inflate excessively because there's a natural "cap"—energy supply. Global energy growth is only 4-5%, and the US has seen just 1% growth over the past decade. Without electricity, you can't build data centers or buy GPUs. However, he is convinced that AI's long-term value is "nearly infinite." Every additional intelligent token can optimize supply chains, finance, logistics, and even address labor shortages. He set a goal: "Reduce the cost and energy consumption of AI by a factor of 10,000," invoking the Jevons paradox where cheaper, more efficient resources actually generate more demand. "The next two to three decades will be a golden age for technologists. We'll solve chemistry, language, new materials, cancer, and lift billions out of poverty. There's no better time to be alive."
Gelsinger's analysis reveals a core proposition: technology companies must be led by people who understand technology. When "financial logic" replaces "technological logic," and when short-term shareholder returns overwhelm long-term manufacturing capacity building, even a once-great company can be quietly left behind by the times.
Comments