Over the past decade, investors have seen: Taiwan Semiconductor Manufacturing (TSM -0.69%) exceed Intel (INTC -0.79%) In terms of process technology.

As a result, almost all major chip manufacturers Apple with Nvidia and many other companies now rely entirely on TSMC for their chip manufacturing needs. In fact, even Intel uses TSMC to produce its latest client computer chips.

But with Intel on the verge of completing its “five nodes in four years” transformation, management is promising that its technology will regain or even surpass TSMC’s next year. The statistic below shows just how big a bet Intel management made.

Intel’s PP&E more than TSMC

The foundry business is an extremely capital intensive business; Chip manufacturers are forced to build large manufacturing facilities and fill them with extremely expensive capital equipment to produce semiconductors. This is why many semiconductor designers have stopped manufacturing their chips altogether.

TSMC surpasses Intel in leading process technology, enabling Intel competitors and TSMC customers to Advanced Micro Devices and with Nvidia ahead, it’s no surprise that TSMC’s investments in property, facilities, and equipment will outpace Intel in late 2020. But what’s surprising today is that Intel has now surpassed TSMC once again after starting to increase its PP&E investments in late 2021. PP&E terms balance (Note: the chart below shows the wrong figure for TSMC from the most recent quarter of 2024, when TSMC recorded $97.1 billion in PP&E).

INTC Net PP&E (Quarterly) data Y Charts

Intel reclaiming the PP&E crown is even more shocking given that Intel’s revenues have been trending in the opposite direction since 2022.

INTC Net PP&E (Quarterly) data Y Charts

Not only does Intel’s PP&E increase contradict its revenue decline, but the company’s PP&E is also on par understated according to these figures. That’s because Intel currently uses TSMC to produce its most advanced computer chips, including its new Lunar Lake and Arrow Lake CPUs, as well as Gaudi accelerators.

So what’s up with Intel’s rising PP&E as its revenue is falling and today’s most advanced computer chips are now produced by TSMC?

Build it and they will come

The answer is Intel CEO Pat Gelsinger’s ambitious plan to catch up with TSMC by 2025. Gelsinger was appointed CEO in February 2021 and soon came up with a plan to reach “five nodes in four years,” a truly rapid pace of innovation. , is becoming a foundry for other chip designers to become a true competitor to TSMC.

The ambitious plan required large investments years before income could theoretically start flowing from those investments. Although the plan started in 2021, the first node in the plan, Intel 7, entered mass production in 2022. The next node, the Intel 4, the first Intel node to use EUV lithography, entered high-volume production in September. 2023. The third node, Intel 3, entered high-volume production in June 2024.

Intel then decided to skip the fourth node in the plan, the 20A node, and outsourced production of the 20A Arrow Lake desktop processor products to TSMC. This was done to speed up the development of the fifth node, 18A, while also saving on capital equipment.

Therefore, most of Intel’s investments will potentially pay off next year. That’s when Intel 3 data center chips will reach high volumes, Intel’s new advanced client chips produced at TSMC will also increase, and 18A will begin producing its first products.

It’s worth noting that Intel’s current revenues mostly reflect products built on only the first two of the five nodes in the plan. in the second quarter conference call with analystsMore than 85% of Intel’s wafer volume still comes from pre-EUV nodes, CFO David Zinsner noted. Therefore, current revenues and margins do not reflect the fruits of the company’s large investments.

All eyes on 18A and 2025

Therefore, mid-to-late 2025 will be the period when returns should reveal themselves. Of course, this assumes a leap of faith that Intel designs good products and will produce them at acceptable efficiencies.

This is a skill set that was lost under previous Intel management, but appears to be back on track under Gelsinger. Gelsinger said in September that 18A had reached a “defect density” of 0.4. While this may not be ready for high-volume production, it would be a solid figure to show that the 18A is on schedule at this stage of the process and should be ready for high volumes by mid-2025, meeting the target laid out four years ago. Intel also said it expects to produce eight “band-in” or chip designs in 18A next year, including both internal and external foundry customer products.

So will this big bet pay off? Intel investors should pray for this to happen, while TSMC and Samsung investors should pray for it not to happen. Only time will tell.