Gelsinger Retires; Intel's Future
Creating a Products division is a win, but challenges ahead for Holthaus.
This morning, Intel announced Pat Gelsinger's retirement.
Pat contributed significantly to Intel’s history as a CPU architect, CTO, and CEO.
As CEO, Pat’s efforts to regain leading-edge semiconductor manufacturing leadership will forever be an important chapter in American semiconductor manufacturing history.
Unbundling
Pat’s Five Nodes in Four Years (5N4Y) plan to regain manufacturing leadership eventually led to an internal unbundling of Intel Foundry to create better financial transparency; Intel (the IDM) split off Foundry into an “independent subsidiary inside of Intel”1. Foundry’s disaggregation was strategically sound and aligned with the 5N4Y vision.
However, Intel's revenue-generating products business floundered under Gelsinger, missing the AI accelerator boom and losing datacenter CPU market share. Most troubling, Pat failed to articulate a compelling vision for Intel’s products over the next 3-5 years, pinning Intel’s product hopes on an AI PC super cycle—a mirage that has yet to manifest2.
Put simply: Pat succeeded as Chief Manufacturing Officer but failed as Chief Product Officer.
Today’s announcement of a new Intel Products Group with Michelle (MJ) Johnston Holthaus as CEO is an important step toward filling that gap.
As Holthaus’ updated bio articulates,
Michelle Johnston Holthaus is interim co-chief executive officer and chief executive officer of Intel Products at Intel Corporation. As CEO of Intel Products, she is responsible for a group that encompasses the company’s Client Computing Group (CCG), Data Center and AI Group and Network and Edge Group.
Holthaus, previously the GM of Intel’s most stable business and enduring brand (client CPUs), fills the Chief Product Officer role that Gelsinger failed to fulfill.
At the same time, the board has officially unbundled “Intel the IDM” into two divisions: a fabless design group (Products) and a fab (Foundry).
Is the board setting the stage for a Foundry spin-out?
Recently, Intel’s 8-K called out a hypothetical “Intel Foundry Corporation” in the U.S. government subsidies agreement:
The Direct Funding Agreement contains restrictions on certain “change of control” transactions: … (ii) Intel ceasing to own at least 50.1% of the ownership of or voting rights with respect to Intel Foundry if separated into a new legal entity (“Intel Foundry Corporation”) so long as Intel Foundry Corporation remains a private company; (iii) if Intel Foundry Corporation becomes a public company, third party acquisition of 35% or more of the ownership of or voting rights with respect to Intel Foundry Corporation at any time Intel is not its largest shareholder; ….
Some suggest a misconception that these terms prohibit Intel from spinning off its Foundry division; this is not the case. Intel can execute such a spin-off, but it has to tick some regulatory boxes which can include Department of Commerce approval.
This simply signals that the U.S. government has a vested interest in Intel Foundry’s stability and prosperity.
Product Rebirth
Pat was correct in spending as much time as possible on 5N4Y; Intel was caught in a death spiral as an IDM with an uncompetitive fab, and catching up to the leading edge of semiconductor manufacturing is one of the most challenging, most unforgiving pursuits in the industrial world.
Yet Intel’s Products group needed the CEO’s full attention, too. Pat couldn’t give 100% attention to both Foundry and Products; who could? (OK, other than Elon).
With Holthaus assuming the CEO role at Intel Products, Intel now has a singular, dedicated leader focused on the product lineup and P&L accountability.
MJ can focus entirely on Products—something that’s been long overdue—and use her new CEO status to drive the necessary, albeit difficult, changes. For example, is the Product Group as lean as it needs to be?
Yet her promotion is no silver bullet. Holthaus has her work cut out to articulate a compelling vision for Intel Products. In this AI world, what’s Holthaus’ 3-5 year vision where Intel Products’ revenue actually grows?
After all, CPUs alone aren’t going to save Intel. This leads me to my next point: managing a legacy x86 CPU division is different from what Intel Products now needs, which is someone who can construct a product portfolio with strategic bets on the future.
I also can’t help but notice that most other major fabless semiconductor companies, such as Nvidia, AMD, Broadcom, and Qualcomm, are led by former engineers. Can Holthaus and Intel CTO Greg Lavender hang with Lisa Su and Mark Papermaster?
Foundry
Given that Gelsinger, who spent a lot of cycles on Foundry, is out – I wouldn’t be surprised if recently hired Intel Foundry leader Kevin O’Buckley is eventually promoted to CEO of Intel Foundry. While O’Buckley is fully dedicated to ensuring 18A lands perfectly, naming him as Foundry CEO sends the message that Product and Foundry are equally important and have the leadership they need.
Yet O’Buckley still has the same problem today that he had yesterday: Intel Foundry is unprofitable, and its principal patron’s coffers continue to dwindle. At least this patron now has clear accountability (Holthauser). But O’Buckley still must deliver 18A perfectly, and meanwhile, Holthauser can’t miss a beat either. In addition, Intel’s board needs to conduct a CEO search. That’s a lot of spinning plates…
Moreover, today’s news doesn’t solve Foundry’s other problems like customer trust and incentives alignment. When the rubber hits the road, O’Buckley must still prioritize his patron over other customers. And don’t forget, she’s a direct competitor to some of his prospective customers.
The only solution is for Intel’s board to complete the unbundling of Product and Foundry.
Take heart: Intel’s org chart is shaping up perfectly for a spin-out into two separate companies:
Could Intel Foundry Corporation be on the brink of becoming a reality? Pat was against this level of unbundling, but now he’s no longer with Intel. This might be why.
Of course, Foundry can’t stand alone yet. But Intel Foundry could get by with a little help from its friends.
