facebook rss twitter

TSMC market value casts a shadow over Intel

by Mark Tyson on 21 October 2019, 11:21


Quick Link: HEXUS.net/qaeexn

Add to My Vault: x

Last Thursday Taiwan Semiconductor Manufacturing Company's market cap surged past that of Intel, reports Barron's. The contract chip manufacturer published better than expected Q3 earnings and showed its ambition by significantly raising its capital expenditure plans by US$4bn (to approx $15bn) to meet "very strong" demand for 7nm chips.

TSMC has been gaining steadily on Intel since around 2015. Graph from Reuters this April.

Breaking down its current figures, TSMC's Q3 revenue was $9.40bn, which is an improvement of >10 per cent YoY, and >21 per cent QoQ. Its advanced 7nm chips accounted for 17 per cent of wafer revenue, its 16nm output for 22 per cent, and 10nm for 2 per cent. "Our third quarter business benefited from new product launches both in premium smartphones and high performance computing applications using TSMC's industry-leading 7-nanometer technology," said Wendell Huang, VP and Chief Financial Officer of TSMC. "We expect the strength of demand for our 7-nanometer technology will continue, driven by high-end smartphones, initial 5G deployment and HPC-related applications."

As mentioned in the intro, capital expenditure plans have been ramped up from $10-11bn for the year to $14-15bn. Of the extra $4bn or so, $1.5bn is allocated to bulking up 7nm capacity and $2.5bn to add 5nm capacity.

Barron's garnered some interesting quotes from a report by Bernstein analyst Stacy Rasgon with regard to TSMC's performance and its place in the competitive chip-making landscape. "Intel's process disadvantage is getting worse," opined Rasgon. "While Intel will theoretically have 7nm parts out sometime in 2021...we believe it likely that CPU and other mainstream 7nm parts will not be on the road map until late that year, if not sometime in 2022." Then, by that time, TSMC will be a major 3nm producer.

Intel closing the process gap seems like an insurmountable challenge as things stand. It can hope TSMC drops the ball, but Intel is equally capable of mishaps and missteps.

HEXUS Forums :: 10 Comments

Login with Forum Account

Don't have an account? Register today!
I doubt TSMC will be a major 3nm producer by 2022. 7nm+ will be in full swing in 2020. In 2021 we will see 5nm node on several products and in 2022 also. So my estimation is that TSMC will have 3nm products out on the market sometime in 2023. Intel needs to do pull a magic trick to stay competitive in this landscape in the next few years.
As you can see here, Intel says that current 10nm process (which I suppose is called 10nm+) is the same in perf as 14nm++, which isn't the case really as we can see on notebooks with Ice Lake vs Whiskey Lake. They both good battery life, but Ice Lake cannot exceed 4Ghz without starting a fire. So if 10nm++ is 2020, 7nm is 2021, 7nm+ is 2022 and 7nm++/5nm is 2023, then yeah they will still lag behind.
Their only solution now is to jump to 7nm next year, but I guess that close to impossible given they weren't able to fix the 10nm fiasco after 4 years of delay.
why not just pay more salary and suck up the TSMC engineers
why not just pay more salary and suck up the TSMC engineers

It's not just the engineers, it's the investment TSMC has in fabrication plants that's a generation or two ahead of Intel.

I just hope Intel don't buy TSMC - it's good for us as consumners, if there's a fierce competition raging (at least it should be if Intel ever get the memo!).
why not just pay more salary and suck up the TSMC engineers

Who says they haven't tried? Intel have a pretty poor reputation for how they treat their employees, so I doubt they would get many if any takers.
IIRC Intel were outsourcing to TSMC, I didn't follow it to see how it played out though. Has it made a difference?