The Winners and Losers of Chipmaking Innovation - 7investing 7investing
Stock Tips Mobile Menu Dropdown Icon

The Winners and Losers of Chipmaking Innovation

Intel is going all-in on introducing its newest chipmaking architecture. Who wins and who loses when it comes to this high-speed innovation race?

February 5, 2024

When it comes to semiconductor manufacturing, the Top Dog takes the spoils.

The chip manufacturing facilities (“fabs”) that can manufacture the smallest and most efficient transistor architectures are the ones who win the most lucrative contracts. And that’s because customers need those most-densely-packed transistors in order to perform computationally-heavy applications.

For decades, Taiwan Semiconductor (NYSE: TSM) has worn the crown as the king of semiconductor manufacturing. Its independent foundry model — where it manufactures entirely for others and not at all for internal purposes — has allowed it to fill capacity with the highest-margin business. Its steady cash flows have allowed it to invest ahead of the curve, especially when it came to introducing new process technologies or purchasing new equipment.

However, TSMC has curiously ceded the cutting-edge innovator role Intel (Nasdaq: INTC) when it comes to embracing the latest-and-greatest lithography technology.

ASML (Nasdaq: ASML) — the Dutch lithography monopoly — has spent a decade to develop its new high numerical aperture, extreme ultraviolet lithography (“high NA EUV”). High NA EUV will allow transistors to be printed and etched that are smaller than two nanometers in length. To put that into perspective, the average human hair is around 50 micrometers; which is 25,000 times larger than these new transistors are being created!

Nano-sized transistors are absolutely necessary for the new wave of artificial intelligence. Chip designers such as NVIDIA (Nasdaq: NVDA) and AMD (Nasdaq: AMD) — and now Apple (Nasdaq: AAPL) and Microsoft (Nasdaq: MSFT) too — have historically worked with TSMC as the only manufacturer who was even capable of producing at the specifications they demanded.

But is that soon about to change?

Intel is tired of lagging behind TSMC in this chip manufacturing race. Its CEO Pat Gelsinger (who even worked alongside its founder Gordon Moore) has committed to overtaking Taiwan Semi by 2025 when it comes to manufacturing leadership. Intel plans to roll out its A18 process — capable of manufacturing 1.8 nanometer transistors with a Gate-All-Around architecture — by the end of the calendar year. Intel will be using this not only to manufacture its own processors, but to woo America’s largest tech companies as well.

Interestingly, Taiwan Semi is heading in a different direction and is in no hurry to deploy ASML’s new high-NA EUV technology. TSMC is transitioning away from its FinFET process to a new nanosheet architecture. This new approach will also allow TSMC to compete for sub-2-nanometer chips, which it hopes to introduce by 2025. The nanosheet architecture does not currently require high-NA EUV equipment or technology.

The 7investing Key Takeaway(s)

For investors, there are several takeaways related to the fast-moving world of chipmaking technology.

Firstly, this is good news for Intel.

The semiconductor manufacturing industry is capital intensive, and a state-of-the-art new fab can be very expensive. Intel made some painful but important capital allocation moves this past year — selling its Data Center Solutions group and then slashing its dividend — in order to focus more resources on its manufacturing capabilities. It is currently investing $20 billion in its new Ohio mega-fab, and it could invest up to a whopping $100 billion (which is government subsidized).

These were necessary changes. Intel’s entire culture has changed, and its internal divisions are now competing for manufacturing capacity and resources. Investors should applaud the move. Intel’s Foundry division grew 63% in the fourth quarter and it generated $4.6 billion in operating cash flow. Capacity utilization will be key to bringing Intel back once again to profitability.

Secondly, this is GREAT news for Applied Materials (Nasdaq: AMAT).

Intel’s new A18 process isn’t just introducing new lithography technology. It’s also introducing new equipment from Applied Materials for pattern shaping.

Pattern shaping involves changing the shape of the small holes that interconnect the layers of a chip, from round circles to ellipses. This innovation allows fabs to maximize the density of transistors on a chip, without adding significantly more costs.

Intel is all for this improved efficiency. A recent Tom’s Hardware post stated that “Intel plans to insert pattern shaping starting from 20A and then High-NA EUV starting from a post-18A node, enabling the company to reduce the complexity of its process flow and avoid the usage of EUV double-patterning.

The new approach will require Intel to buy Applied’s new Centura Sculpta machinery. But it will also eliminate the expensive and inefficient step of lithography known as multi-patterning. Intel is already placing orders with Applied Materials, and that will be a nice boost to the company’s top line.

Thirdly, this could be moderately bad news for ASML.

Let’s be clear: ASML is an innovative company. Its lithography machines will always be very important components as Moore’s Law marches onward.

Yet it also might not be selling as many of its high-NA EUV machines as it initially was hoping. Intel is looking for cost-savings through pattern shaping, and TSMC reportedly won’t adopt high-NA until at least 2030. The company finds itself in a bit of an awkward situation.

And lastly, this could potentially be really bad news for TSMC.

Taiwan Semi is really trying to expand outside of Taiwan, to counteract the rising geopolitical risk of a Chinese invasion. However, it’s also delaying the timeline of its Arizona-based fabs due to it having trouble finding enough talented employees to man its operations.

TSMC needs to fix this issue immediately. If it doesn’t, it is leaving the door wide open for Intel to win over Apple as its newest customer.

Chip architecture changes sound exciting in presentations and they look great on paper. But they’re also incredibly hard to pull off — especially once the technical issues like manufacturing yields, costs, and efficiency come into play.

Intel, Taiwan Semi, Applied Materials are all redlining on innovation when it comes to the semiconductor manufacturing race. Investors can look forward to things continuing to run full-speed-ahead in 2024.

Join 7investing's Free Email List

Already a 7investing member? Log in here.