Semiconductor-giant Intel (INTC +7.06%) has its sights set on slashing annual costs by as much as $10 billion by the end of 2025. Until now, it was somewhat unclear where those cost savings would come from. The company has done some minor layoffs, temporarily cut salaries, and exited a slew of non-core businesses, but all of that isn't nearly enough for it to hit its target.
Intel announced the restructuring of its manufacturing operations on Wednesday as part of its plan to build a world-class foundry business to challenge market leader TSMC. And now, the pieces of the puzzle have finally fallen into place.
NASDAQ: INTC
Key Data Points
An opaque cost structure
Intel designs and manufactures its own chips. There are costs associated with both processes, but traditionally, the manufacturing side of the business hasn't been operated as a stand-alone unit. In the past, when Intel faced minimal competition in its core PC and server-chip markets and its manufacturing technology was the best in the industry, this setup created no meaningful problems.
The industry has changed. Intel faces a resurgent AMD in the PC and server-chip markets, and TSMC is now the clear leader in semiconductor manufacturing. Intel is investing tens of billions of dollars to build a foundry business of its own and reclaim the manufacturing crown, but its old structure is holding it back. Simply taking all manufacturing costs and spreading them among business units in an opaque way has created enormous inefficiencies.
One example of such an inefficiency is the company's heavy use of running expedited wafers through its manufacturing facilities. At a third-party foundry, there would be a significant cost in doing this.
Expediting the production of wafers hurts the productivity of the factory, but with Intel's old setup, business units wouldn't pay for this directly. There was no incentive not to expedite wafers because there was no real downside at the business-unit level.
Another example is test times. Intel found that the time its manufacturing arm spent on chip testing was double or triple that of competitors. Again, there was no incentive for business units to make design choices that minimized testing because those business units weren't paying for testing directly.
Finding billions in cost savings
Intel is aiming to fix this problem with a sweeping reorganization of its operating model. Beginning in the first quarter of 2024, the manufacturing operation will be treated as a stand-alone unit, complete with a reportable profit and loss statement. Other business units will interact with the manufacturing unit similarly to how they'd interact with a third-party foundry, paying market-based rates for all services.
On the manufacturing side, this new structure will provide the clarity necessary to operate efficiently. And for other business units that tap into the manufacturing operation, there will be strong incentives to avoid wasteful practices, like expediting an excessive number of wafers. The costs associated with those actions will show up directly in the business unit's results, instead of being opaquely distributed across the company.
The potential for cost savings by switching to this new structure is enormous. Intel estimates that as much as $1 billion can be saved annually from the reduction in expedited wafers. Another $500 million in costs can be cut by reducing testing times to industry norms, and an additional $1 billion of savings can come from reducing the number of wafer steppings, or physical iterations of a product's design.
This change will require a shift in culture, which is never easy and almost always comes with unintended downsides. But it's well worth doing. Billions in annual savings are on the table just by creating the right incentives among the company's various business units.
Intel will need to find more costs to cut to reach its goal of $8 billion to $10 billion in annual cost savings by 2025. But this restructuring is a big step toward that goal.
