It’s no secret that Intel has shone more with a lack of strategic direction than with a pioneering technological spirit in recent years. Following the departure of Pat Gelsinger, Lip-Bu Tan has now taken over the helm – and the new CEO doesn’t seem to be the type to sit in the third row and just sketch PowerPoint slides. Instead, the AI division has been declared a top priority – in the truest sense of the word.
Flat hierarchy, tough course: Tan turns things upside down
According to a memo that has surfaced at Reuters, Tan has “flattened” the management structure. A term that usually means roughly the same thing in corporate communications: fewer people between the problem and the responsibility. DC (Data Center) and AI (Artificial Intelligence) now report directly to the CEO. Michelle Johnston Holthaus, previously responsible for the product portfolio, will take on other areas in future. Details? Later. Much later. In addition, Sachin Katti, formerly a network expert and now deeply rooted in technology at Intel, will become CTO and “AI Chief”. Two hats for one head – a clear sign that Intel finally wants to pick up speed again. Or has to.
Fun instead of games: why Intel’s AI chips are not (yet) fun
The sad truth: While NVIDIA makes more margin with each GPU than others with an entire CPU lineup, and AMD at least delivers with MI300 fodder for data centers, Intel is looking pretty pale. The Gaudi accelerators – purchased years ago via Habana Labs – eke out an existence somewhere between “also available” and “nobody asks for it”. Technically not necessarily bad, but without a significant software base and without OEM backing. As is so often the case with Intel, the timing was unfortunate. While others were already chanting the “AI-first” mantra, they were still hard at work on Meteor Lake in Santa Clara. The AI train had already set off, while Intel was still looking for the timetable.
Geopolitics as a spoilsport
As if that wasn’t enough, the US government is now also banning the export of Gaudi chips to China. Reason: sensitive AI technology. Result: one more market segment less. Not an ideal starting position for a company that relies on scaling. The already limited demand is additionally neutered by regulatory hurdles. The plan to finally catch up with Gaudi 3 (codenamed “Jaguar Shores”) is therefore likely to be even more difficult to implement – even if the technology is actually competitive this time. And that remains to be proven.
Tan’s management style: roll up your sleeves instead of straightening your tie
Tan himself has left no doubt in his internal communication that he wants to get closer: “Roll up your sleeves” and talk directly to engineering. This sounds like a start-up mentality, but it applies to a company with 130,000 employees and a hierarchy that is more reminiscent of a civil service than Silicon Valley dynamics. Whether this approach is more than just symbolic action remains to be seen. After all, when the CEO talks to the developers in person, it sends a signal – even to those who have had to drag their ideas through five departments up to now.
Conclusion: now or never – but with substance, please
The AI realignment under Lip-Bu Tan could be an important step – or another chapter in Intel’s “strategy through restructuring” chronicle. The market will have no patience. NVIDIA dominates, AMD is moving into position, start-ups like Tenstorrent are lurking for any gap. Tan has to deliver – not just new titles and reporting levels, but real products, real performance, real relevance. Otherwise, the new organization chart will be just another PDF in the archive of failed Intel attempts. Intel can do AI. Theoretically. Now the company has to show that it can do it in practice.
Source: Reuters
Bisher keine Kommentare
Kommentar
Lade neue Kommentare
Artikel-Butler
Alle Kommentare lesen unter igor´sLAB Community →