TSMC Vs. Intel: The New King Of The Hill

Investment Thesis

Although I have discussed both Intel (NASDAQ:INTC) and (to lesser extent) TSMC (NYSE:TSM) previously, it’s been a while since I have compared both. In line with the overall market, both stocks have declined quite substantially from their highs. For TSMC this means, combined with its continued growth, that it has caught up to its previously quite elevated valuation. For Intel, even the roughly 50% drop still hasn’t been enough to fully offset the substantially decreased earnings estimates as revenue has fallen and investments have decreased.

While it may be tempting, as Buffett’s Berkshire Hathaway (BRK.A)(BRK.B) has done, to conclude that TSMC remains the top dog of semiconductor manufacturing, I would nevertheless argue that Intel’s prospects going forward are just as bright, as substantial leverage should become visible once the current investments start to pay off. Meanwhile, TSMC is visibly struggling with its latest nodes.

So while a case for both companies could be made at their current valuation, I’d buy Intel over TSMC as the new king of the hill for the next decade.

Competitive landscape

This article has been inspired in part by a similar recent comparison by another contributor. While the author laid out many of the most relevant facts, what I differ in is in the interpretation of those. To wit, as has been a consistent theme throughout my coverage of both companies, the topic of process leadership.

While it is true that 10nm had been delayed for many years, ultimately this is not relevant anymore as Moore’s Law has continued to shift the goalposts. As such, 10nm is just as relevant to this discussion as who won World War II.

So when it comes to who is best positioned for semiconductor leadership going forward, my analysis has indicated it is Intel, not TSMC. The reason for this is simply that Intel has its 2nm node (called 20A, followed by 18A six months later) lined up for production to start in the first half of 2024, which compares to TSMC’s equivalent node (called N2) which is scheduled for the second half of 2025, a non-insignificant 12- to 18-month lag compared to Intel.

Moreover, Intel said in 2019 it was targeting a 2x shrink, while TSMC’s official disclosure is for a shrink of “>1.1x”, which suggests Intel’s 18A could outperform TSMC’s N2 despite being a year earlier. Indeed, my current estimate is that 18A will have an over 40% higher density (half a node) than N2.

Now, as always, some people might be skeptical about this, so let’s recap some of the reasons why this prognosis should be reliable.

First, even in the face of its many delays, Intel has actually maintained its status as the industry’s innovation powerhouse. The main issue was not with the technology itself, but that it contained too many defects to go into production economically. One of the reasons for this was that Intel was not ready for EUV lithography (because the original 10nm schedule was years ahead of when EUV was eventually ready). In addition, Intel has parallel development teams. So while the Intel 4/3 team stumbled upon a hiccup, the Intel 20/18A team just continued to move forward.

In addition, Intel will introduce two key brand-new technologies in 20A, namely RibbonFET (gate-all-around FET) and PowerVia (backside power delivery network). Investors can be assured that these technologies have not just been put on the roadmap a few quarters ago, since in reality such ground-breaking innovations usually take at least a decade to go from research to production. As of October 2022, Intel stated that these nodes remain on track, with the first full-blown test chips (both internal and from a foundry customer) now in the fab.

Lastly, there is one more reason that Intel’s current nodes such as Intel 7 are irrelevant, which is that onboarding new foundry customers literally takes years, as that is the time it takes for the development of new chips. Note that IFS (Intel Foundry Services) was only created in early 2021, which puts the earliest volume production for those initial customers likely in 2024-25. But that is exactly the timeframe when 20/18A will become available, which proves the point that earlier nodes are hardly relevant anymore; new potential foundry customers will be looking at Intel’s offerings for the 2025+ timeframe.

As a side note, Intel mainly expects foundry customers to adopt 18A, as 20A will be more of an internal, lower-volume node like Intel 4, as part of Intel’s new Tick-Tock process development methodology.

Of course, some customers will likely have more or less stringent requirements with regards to the milestones they would want to see before increasing their adoption of Intel’s foundry services. For example, if a new customer like MediaTek, which Intel announced is adopting Intel 16 initially, would wait until its first chips are in the market before committing to new (perhaps more leading edge) projects, then that would take many years. Perhaps closer to reality, though, as Pat Gelsinger has explained, the various test chips on 18A that will come out of the fab over the coming quarters will serve as proof points for potential customers as they are evaluating where to get their future chips manufactured.

Pat Gelsinger: Yes. Thank you. And on 20A and 18A, they go to RibbonFET, as you say. And Intel has driven every major transistor, right, in the volume production for the last 35 years. So the idea that we’re the ones who are going to drive this major new transistor structure into production is something that we’re pretty committed to be a driver for 20A, as you said, on track, on schedule. We expect 20A will primarily be an internal node, not one that we have a lot of external foundry customers for the external foundry chipset or tape-outs are largely associated with 18A.

And a very typical process for a foundry customer will be “give me a test chip of my circuits on your process.” and that’s exactly what we tape out. The first one this quarter. We’ll have several more in the pipeline. So now we’re taping out not only our test chips for 18A, but our foundry customer test chips for 18A, and that’s a pretty critical milestone when they see the results of the silicon for them making a volume decision for a foundry customer.

So we’re exactly on the time line that I described earlier for those tape-outs and those decisions. So as they start to see the silicon results, which we think are going to be very promising we think that will be a key step to them making major foundry decisions. And overall, this just affirms our five nodes in four years. We’re making the investments. We’re seeing good progress to get back to process technology leadership, which for Intel is a tide that raises all boats in the company. It makes our products better. It establishes our new business areas, positions us in a very profound way for foundry

Conclusion

Intel’s 18A node represents the first possible interception point for potential foundry customers to become an early adopter of a new Intel node since the IFS business was created, and therefore this node is pretty much the only sensible node to use as comparison to evaluate process leadership. Using any other node for this purpose would be a straw man at best.

So in that regard, based on density projections disclosed by Intel and TSMC, my current estimate is that 18A will be half a node denser than TSMC’s N2, while also being a full year earlier to market. Unless any changes in the schedules were to occur, this hence leaves no question of who will be able to claim process leadership going forward: not TSMC, but Intel.

One example of how this could play out in the market is if Qualcomm (QCOM) would launch an 18A SoC in 2025, whereas Apple’s 2025 products will have to use some variant of TSMC’s N3, with an N2 iPhone only becoming possible in 2026.

Valuation

Even after the substantial rally since the Berkshire Hathaway investment in TSMC became public, TSMC is currently still valued at just a double-digit P/E, with the stock price down significantly. Clearly, in a more bullish market, TSMC could easily sport a valuation (multiple) in the 20s or even 30s, even before considering any future growth. It is projected that the foundry market will continue to grow at a decent clip (Intel’s projection at Investor Meeting 2022 was that it would grow from $100B to $170B or so by 2030). In other words, even if TSMC loses some market share to newcomer Intel, unless IFS becomes more successful than any single person (including Pat Gelsinger) currently anticipates, then TSMC is unlikely to deliver negative alpha going forward.

In contrast, Intel’s forward P/E of nearly 15x is actually a bit more expensive than TSMC. However, this comes with the caveat that Intel has lost a substantial amount of leverage due to the current decline in revenue and the ongoing investments to catch up. However, these investments are likely to pay off given the trends with regards to process leadership as discussed above. This means Intel’s earnings could multiple once it starts to grow revenue again.

For example, management openly stated during the last earnings call that it aspires for Intel to deliver industry-leading metrics such as gross margin (60%+) and operating margin (40%+). Clearly, Intel is currently delivering far below those benchmarks, leaving ample upside if it succeeds in what Pat Gelsinger has previously called the greatest turnaround ever.

Investor Takeaway

Despite their similar (currently quite low) P/E valuations, Intel and TSMC are two companies with two quite different profiles in terms of several metrics such as gross margin, growth, and operating margin.

At first sight, the investment case for TSMC sounds reasonable enough: it is the foundry industry leader (both in technology and market share), and one may be tempted to fall into a false sense of security by assuming this will remain the case going forward. Hence, continued (industry) growth should make for a safe investment.

While I do not necessarily disagree with this reasoning, the more alluring investment going forward would actually be Intel. First and foremost, given the deflated gross and operating margins, Intel is arguably the company with the greatest opportunity to multiply its earnings (and with that its stock price). Secondly, as discussed Intel is currently unambiguously in pole position to take over process leadership from TSMC in 2025. Once this happens, because TSMC’s N2 is unlikely to fully catch up to 18A, Intel will then likely (comfortably) hold this position at least for the rest of the decade.

Since 18A is also the first node that lines up with the development cycle of chips, after onboarding the first customers in 2021, this means IFS will enjoy a leadership position out of the gate, which bodes well for this business’ prospects, which will come at the detriment of either Samsung or TSMC as its only competitors at the leading edge.

Ergo, Intel is (will be) the new king of the hill.