| | | D2D - Intel and the Foundry State of Play
Subscribe to D2D Newsletter
Long/ interesting article and worth the read (IMO) .
Executive summary and conclusion below (Derivative Impacts)
Executive Summary
Intel has arrived at the crossroads. It needs to decide if it will continue to manufacture semiconductors. To stay on the path of Moore’s Law, we estimate it needs $15 billion to $25 billion, and it is not clear who will supply that. Instead, if they choose to abandon manufacturing they need to move quickly to save the Product side of the company.
As much as most observers have already written Intel off, we think it is critical for the semiconductor industry, that Intel stay in the game. Without Intel, and Samsung Foundry on shaky ground, there is a non-zero chance that TSMC becomes a true monopoly. And like any monopolist it will eventually raise prices to the point where many customers will choose to delay moving to advanced nodes, which will slow down the entire technology industry.
Derivative Impacts
A common argument now holds that there is no commercial justification for Intel to remain in manufacturing. TSMC has out-competed them, and no one seems worse off for it. This is the logic that has led the company to seek out government support. Absent a commercial justification for its existence, surely it has a national security justification.
While we certainly see the importance of Intel for the US strategically, we think there is a massive commercial need for its existence as well. As it stands now, TSMC has an effective monopoly on advanced semis manufacturing. “Effective” in the sense that both Intel and Samsung are still notional competitors. That, coupled with a strong customer-centric culture, have kept it from really flexing its competitive leverage. But that will change if Intel exits the business entirely and TSMC becomes an actual monopoly. Samsung will still be out there, but given that TSMC has been out-competing them for a decade and Samsung’s ambivalence about external customers, we think they will not be able to effectively curb TSMC’s advantages.
For several years, TSMC’s shareholders have been pressing the company to raise prices. They question why Nvidia – who is entirely dependent on TSMC – has much higher gross margins than TSMC. Absent a true competitor, we think it is only a matter of time before TSMC would raise prices.
The press is already reporting that wafer prices for TSMC’s upcoming 2nm process could reach as high as $45,000-$50,000/wafer, considerably ahead of the current $25,000 - $30,000 price. A TSMC price of $50,000 would fundamentally re-order the semiconductor industry. At those prices, only a handful of companies could afford ordering leading-edge products. Even large companies like Apple and Qualcomm would face significant gross margin pressure on their own products. We estimate that at $47,500/wafer Qualcomm’s gross margins would go to zero. These companies will still order chips from TSMC, but they will stay on older nodes much longer. Consumers would notice that new devices offer few improvements over previous generations. Only companies with the richest gross margins could afford to stay at the leading edge, effectively shutting out their competitors, and wreaking havoc with semis start-ups. This would be the effective end of Moore’s Law.
This will also ripple through the equipment side of the industry. We are already seeing Intel’s move away from past capex levels hurt earnings at companies like ASML and (probably) Applied Materials. TSMC would not only be the monopoly supplier, but also the monopsonist customer for the wafer fabrication equipment (WFE) industry.
In fairness, this is the worst case scenario. We imagine the US government and Intel will probably reach some kind of deal. And Samsung could very well turn itself around. But there is still a significant probability that both Intel and Samsung exit the business leading to the gloomy scenario above, and it is important that the broader industry recognize how much it will affect all of us. |
|
