TSMC: Weak 2019 Demand, but 5nm Set for 2020 Volume Manufacturing

There’ve been a lot of nervous eyes on TSMC of late. Apple’s iPhone sales shortfall bit the company hard earlier this month after it announced an expected 14 percent drop in Q1 2019 revenue relative to the same period in 2018 and that it expected demand to remain weak in the first half of 2019. Following on the heels of reports that TSMC’s 7nm node is running under capacity, the potential for a general semiconductor slowdown has had investors nervously eyeing the company’s sales projections.

At its Q1 conference call this week, TSMC announced some mixed news on that front, but it signaled confidence in its 5nm technology development roadmap. 2019 revenue is only expected to increase by 1-3 percent, compared with 6.5 percent in 2018, as the company clears inventory congestion. There are claims of general slowdowns throughout the industry — DigiTimes reports that rival foundries like UMC, SMIC, Vanguard, and Powerchip have all been cutting their price quotes in a bid to pick up additional business, with prices falling by as much as 20 percent.

SemiWiki reports the following comments from the President and co-CEO of TSMC, CC Wei:

Our N5 technology development is well on track, with customer tape-out schedule for first half 2019 and volume production ramp in first half 2020. We are already in preparation for N5’s ramp. All applications that are using 7-nanometer today will adopt 5-nanometer. In addition, we are expecting the customer product portfolio at N5 and see expanding addressable market opportunities. We expect more applications in HPC to adopt N5. Thus we are confident that N5 will also be a large and long-lasting node for TSMC.

It’s interesting to compare TSMC’s revenue splits now with a few years back. First, here’s revenue for full year 2018, split by process node:

Now, here’s the same revenue split (with more granularity at the per-node level) from back in Q1 2015:

In Q1 2015, TSMC’s leading edge node (20nm) was producing 16 percent of its revenue, while its mainstream 28nm node accounted for 30 percent. The remaining 54 percent of revenue was split into older nodes. Now, we see a different pattern. 10nm has fallen to a very small percentage of the overall pie, reflecting its status as a shorter-lived node for the company. But despite being roughly the same age as TSMC’s 28nm node was in Q1 2015, its 16/20nm node combined accounts for less overall revenue, while the share from 7nm is proportionally greater. (TSMC began volume production of 20nm chips in June 2014 and 16nm in August 2015. Volume production of 7nm parts started in July 2018).

All of this dovetails with TSMC’s CEO comments about its customer base for 7nm having already committed to 5nm production. The suggestion here is that the high-end customers who are continuing to adopt the latest and greatest in semiconductor technology have committed to doing so at another node, at least for now. Comments from GlobalFoundries earlier this year suggested that the absolute number of 7nm customers had become a problem for sustaining operations as an independent foundry competing at the cutting-edge. Samsung’s own 7nm announcements have also been relatively light on customers, though rumors have suggested that Nvidia is working with the company on that node and IBM has inked an agreement to build future POWER chips with the Korean foundry.

The major challenge at 5nm will be EUV integration. 5nm is expected to be a key inflection point for the new manufacturing technology. At 7nm, EUV will only be used for contacts and vias, which can be manufactured without the need for a pellicle. At 5nm, EUV will be used more extensively and for manufacturing steps that require a pellicle, a thin, transparent layer that keeps dust particles from landing on a mask. Pellicle design has been a sticking point for EUV thus far, with the best public demonstrations not yet hitting the minimum EUV transparency foundries require. Presumably, TSMC believes it has this problem solved or that solutions will be online by next year.

Continue reading

Activist Firm Urges Intel to ‘Explore Alternatives’ to Manufacturing Its Own Chips

Intel is facing calls from an activist investor to explore strategic alternatives and potential for a spinoff or divestment of previous acquisitions.

Intel Urged to ‘Explore Alternatives’ to Manufacturing Its Own Chips

Intel is facing calls from an activist investor to explore strategic alternatives and potential for a spinoff or divestment of previous acquisitions.

Intel May Postpone Manufacturing Decision, Emphasizes Beating Apple

Intel may postpone its decision on whether to build 7nm CPUs in its own fabs or to partner with an outside foundry for future leading-edge nodes. Also, Pat Gelsinger has some harsh words for Apple.

EU Wants 20 Percent of Semiconductor Manufacturing by 2030

The EU may unveil a plan to boost its share of leading-edge semiconductor manufacturing by up to 20 percent of the market by 2030.