Intel’s Former Glory: Three CEOs and IDM 2.0

Intel’s Former Glory In 2000, Intel’s market capitalization reached US$300 billion, making it the undisputed semiconductor leader. It was then six times TSMC’s and sixty times NVIDIA’s. Now it is about one-eighth of TSMC’s and less than one-thirtieth of NVIDIA’s. Beyond layoffs and delayed fab construction, Intel is considering selling its FPGA division to raise more cash. In recent years, the business that remained profitable and grew rapidly was Altera. Short of cash, Intel has to consider selling its most profitable division. It acquired Altera in an all-cash US$16.7 billion deal in 2015. Ten years on, accounting for inflation, what price would it need to avoid a loss? ----- A Review of Intel’s Three Most Recent CEOs Brian Krzanich (2013–2018) With the PC market shrinking rapidly, the technically trained Krzanich faced the responsibility of transforming Intel and shifted its focus toward the Internet of Things (IoT) and the cloud. He did not believe the lithography machines could achieve economies of scale and passed on ASML’s first-generation extreme ultraviolet (EUV) equipment. Intel repeatedly missed its 10-nanometer milestones, fell behind TSMC and Samsung in advanced processes, and even lost market share to AMD. Robert (Bob) Swan (2019–2021) He inherited an Intel burdened with problems such as the stalled 10-nanometer process, which remained difficult to overcome. During his tenure, Intel’s dominance gradually declined toward parity with AMD, which caught up in some markets. Apple began introducing its own M1 processor and abandoning Intel. OpenAI and Intel even discussed an investment in 2017–2018. But Swan thought generative AI models would be difficult to commercialize in the near term and unprofitable, and the deal fell through. Swan came from finance and was not a technical insider, but he did one thing right: in 2020, he worked hard to reach a partnership with TSMC. Had that cooperation expanded, it would inevitably have developed toward outsourcing all Intel manufacturing to TSMC. Intel might look entirely different today. But opposition within Intel was fierce, and it refused to hand manufacturing to TSMC. Intel still could not resolve its 10-nanometer and more advanced process problems, and Swan was forced to resign after only two years. Pat Gelsinger (2021– The technically trained Gelsinger made a major decision after taking over: aggressively advance ‘IDM 2.0.’ The plan was to invest US$100 billion over the following years and build eight fabs worldwide, delivering five process nodes in four years: Intel 7, Intel 4, Intel 3, Intel 20A, and Intel 18A, starting in 2021. Intel handles design, manufacturing, packaging, and testing in a fully integrated model. Can it overtake an industry trend in which different companies handle those stages? See my earlier discussion of vertical integration’s advantages and disadvantages. What looks powerful can leave every production stage stuck when just one link fails. Let us keep watching. ---- Still, Intel is not entirely out of the fight. Its IC design remains at the global frontier, and the US government stands behind it with enormous orders, tens of billions of dollars in subsidies, and substantial loans. With that backing, even strategic mistakes need not force Intel out of the competition.