Latest / The Pivot Podcast with Fexingo: How Businesses Change Direction and Survive Disruption / How Intel Survived The Manufacturing Crisis
Transcript
- Lucas: If you look at the semiconductor landscape today, specifically around October twenty twenty six, there is one story that defines the tension between legacy dominance and disruptive innovation: Intel's near-collapse and subsequent restructuring. Luna: It feels like every other week there is another headline about Intel losing market share or missing a process node target, but the real story is how they are trying to dig out from under that weight. Lucas: Exactly. For thirty years, the vertical integration model was their superpower. They designed the chip, they built the factory, they sold it to the OEM. It was called IDM, integrated device manufacturer, and it worked until the economics of Moore's Law made it impossible to sustain alone. Luna: Right, because building those advanced fabs costs tens of billions, and if you are not selling chips to everyone else, your return on invested capital tanks compared to a pure-play foundry. Lucas: That is the core friction. By the time Pat Gelsinger returned as CEO in twenty twenty two, Intel had fallen behind TSMC by two full generations of process technology. That gap meant their products were less efficient, hotter, and more expensive, which caused customers like Apple and even Microsoft to look elsewhere. Luna: So when he announced IDM two point zero, he wasn't just tweaking a roadmap, he was fundamentally admitting that the old way of doing things was broken. Lucas: He was admitting it loudly. The strategy had three pillars: make Intel's own factories world-class again, expand those factories to manufacture chips for other companies through Intel Foundry Services, and partner with external foundries for parts of the product stack that don't need leading-edge nodes. Luna: Partnering with competitors sounds counterintuitive for a company that has always prided itself on controlling every step of the value chain. Lucas: It was a massive cultural whiplash. For decades, Intel engineers viewed external manufacturing as a failure. To tell them, hey, we are going to send our non-leading-edge designs to Samsung or GlobalFoundries, required a complete rethinking of corporate pride. Luna: I remember reading about the internal resistance. People felt like it was betrayal, or at least a sign that Intel was giving up its soul. Lucas: It was painful. But Gelsinger knew that without cash flow from external foundry customers, they couldn't afford to build the newfabs in Arizona and Ohio fast enough. The government subsidies from the CHIPS Act helped, but the initial capital commitment was staggering. Luna: Let's talk about that capital. We are seeing headlines about Intel cutting jobs and delaying dividends, yet they are still pouring money into these massive construction projects. Lucas: Yes, the financial discipline has been brutal. They slashed headcount by roughly ten percent in early twenty twenty four to free up cash for CapEx. It was a signal to Wall Street that they were prioritizing survival and infrastructure over short-term earnings per share metrics. Luna: That must have been terrifying for employees who already felt the shadow of layoffs hanging over the division. Lucas: It created a divide between the data center group, which was still profitable, and the foundry arm, which was bleeding cash. Gelsinger had to convince investors that the foundry business could eventually compete with TSMC on margin, not just volume. Luna: And the timeline? When did the actual customer wins start showing up? Lucas: The first major public win came with a European automotive chipmaker using Intel seven nanometer process. It wasn't a billion-dollar deal, but it proved the pipes worked. Then came the announcement with AMD, yes AMD, allowing them to use Intel's packaging technology for certain server chips. Luna: AMD using Intel's packaging? That is huge because advanced packaging is becoming just as important as the transistor density itself. Lucas: Precisely. Chiplets. Instead of making one giant, defect-prone wafer, you make smaller pieces and stitch them together. Intel's Foveros packaging technology is actually leading here, even if their lithography lagged behind. Luna: So the pivot isn't just about catching up on transistors, it's about finding niches where they can be unique, like packaging and high bandwidth memory integration. Lucas: That is a smart angle. While TSMC dominates general logic, Intel is positioning itself as the leader in heterogeneous integration. It plays to their strengths in system-level engineering rather than just pure process shrinking. Luna: But can they really compete with TSMC's scale? TSMC has like eighty percent of the market for advanced nodes. Lucas: They won't catch TSMC in total volume soon. But they don't need to. They need fifty percent of the geopolitical importance. The US government wants a domestic supplier for defense and critical infrastructure. Intel is that supplier. Luna: So this is partly a national security play as much as a business strategy. Lucas: Absolutely. The CHIPS Act guarantees a floor for Intel's revenue. Even if commercial customers hesitate, the government contracts provide stability that allows them to rebuild trust with private sector clients over time. Luna: What about the consumer side? Are regular people seeing any difference in Intel's products right now? Lucas: Mostly in the Arc graphics cards. It was a risky move to enter the GPU market against NVIDIA and AMD, but it forced Intel to innovate in software and driver quality, which bled back into their CPU performance optimization. Luna: I haven't seen many gamers switching to Arc, but I have heard IT managers appreciate the power efficiency gains in the latest Xeon processors. Lucas: That is the quiet win. Enterprise buyers care about total cost of ownership. If Intel's new silicon uses less electricity per teraflop, cloud providers will migrate workloads, even if the raw speed is slightly lower than the competition. Luna: It seems like the narrative has shifted from 'Intel is dead' to 'Intel is learning again.' Is that fair? Lucas: Fairer, yes. The arrogance is gone. You see it in how they talk about partnerships. They used to say 'we do everything better.' Now they say 'we collaborate.' That humility is essential for a turnaround. Luna: Do you think they can pull it off by twenty twenty eight? Lucas: By twenty twenty eight, Intel should be running its new Ohio fab at meaningful capacity. If the yield rates hold, they will finally have the volume credibility to attract bigger foundry clients beyond automotive. Luna: And if the yields fail? Lucas: Then we face a bifurcated chip industry. One tier for advanced logic held by TSMC, and a second tier for mature nodes dominated by Intel and Samsung. It is less glamorous, but perhaps more sustainable. Luna: There is a coffee link in the description if you want to keep this conversation going ad-free. Lucas: Honestly, if today was worth a coffee to you, that is the link — buy me a coffee dot com slash fexingo — that's F. E. X. I. N. G. O. Luna: Thanks for supporting the network, Lucas. Back to the tech. Lucas: The key takeaway for listeners managing complex organizations is that identity crises require identity shifts. Intel stopped trying to be what it was thirty years ago and started building what it needs to be today. Luna: It is a reminder that survival often looks like surrender before it looks like victory. Lucas: Exactly. They surrendered the illusion of total control to gain the reality of long-term relevance. Luna: We will see if the Ohio fabs deliver on that promise next year. Lucas: Watch the quarterly capex reports closely. That is where the truth lives.