Waning consumer demand sees TSMC miss revenue forecasts for first time in two years

[midian182]

What just happened? Even the biggest tech giants aren't immune to the effects of a stuttering global economy and waning consumer demand: the Taiwan Semiconductor Manufacturing Company (TSMC) has just missed quarterly revenue forecasts for the first time in two years.

The global economic crisis brought about by skyrocketing inflation, rising interest rates, and increasingly expensive living costs has seen many consumers rein in their non-essential spending, especially when it comes to expensive tech items.

Samsung, which has often been able to weather the storm during times of economic uncertainty thanks to its broad portfolio, saw its recent quarterly profits fall to an 8-year low. As reported by Bloomberg, TSMC, the world's biggest contract chipmaker, is also feeling the heat. Despite its fourth-quarter revenue increasing 43% to NT$625.5 billion ($20.6 billion), it still fell short of the NT$636 billion analysts expected, marking the first time in two years that the company has missed forecasts.

Reports from last month claimed that utilization rates for TMSC's 7nm and 6nm fabs are expected to fall to 50 percent as clients order fewer wafers, while the 28nm fabs that remain at capacity as of Q3 2022 will start to slow down heading into 2023. The Taiwanese firm is also reportedly renegotiating contracts with customers AMD, Nvidia, and MediaTek, who are cutting orders in the face of falling demand for their products. The Covid-related disruption at Apple supplier Foxconn's Zhengzhou plant has impacted TSMC, too.

The situation doesn't look set to improve in the near term. TSMC is expected to see its revenue decline 15% in the year's first quarter, with another drop to follow in Q2 2023. The long-term outlook is more optimistic, though some analysts believe the company could reduce spending on its expansion plans this year as it did in 2022; TSMC was set to spend $44 billion on upgrading capacity last year, but that figure was reduced to $36 billion.

TSMC is investing billions in its US facilities; its Arizona location will include an upgraded 4nm fab and a 3nm fab for producing the most advanced chips. The Chinese nationalist tabloid Global Times isn't happy about the investment, calling it a "dark turn" in the global semiconductor industry. The publication even claimed the US "tricked" TSMC into building the Arizona fabs and is stealing tech from Taiwan.

Permalink to story.

https://www.techspot.com/news/97218-waning-consumer-demand-sees-tsmc-miss-revenue-forecasts.html

 

[Theinsanegamer]

Gee I wonder if the insane cost is turning people away.....Hmmmmm......

Side note, we recently quoted replacements for our hyper V hosts. The costs have risen from $15k per server to $43k. Double the RAM and SSD storage instead of spinny disks have nearly tripled the cost. Even specs identical to our old servers are over $30k now.

So now instead of buying new servers were going to grab some SSDs, upgrade our servers, and keep them around for another 5-6 years.

 

[yRaz]

Gee I wonder if the insane cost is turning people away.....Hmmmmm......

Side note, we recently quoted replacements for our hyper V hosts. The costs have risen from $15k per server to $43k. Double the RAM and SSD storage instead of spinny disks have nearly tripled the cost. Even specs identical to our old servers are over $30k now.

So now instead of buying new servers were going to grab some SSDs, upgrade our servers, and keep them around for another 5-6 years.

It's like all it takes to solve the e-waste problem is money.....

 

[Burty117]

Gee I wonder if the insane cost is turning people away.....Hmmmmm......

Side note, we recently quoted replacements for our hyper V hosts. The costs have risen from $15k per server to $43k. Double the RAM and SSD storage instead of spinny disks have nearly tripled the cost. Even specs identical to our old servers are over $30k now.

So now instead of buying new servers were going to grab some SSDs, upgrade our servers, and keep them around for another 5-6 years.

Did a few physical hardware replacement projects last year, usually in the £100-120k range for 3-4 hosts and a SAN. Then you gotta add all your licensing on for an extra £40k+

RAID cards were particularly hard to come by at one point. It would seem Microsoft, Amazon have had to grow their datacentres very quickly and eaten a lot of the supply chain, pushing up prices on most server parts.

Either that or there is something going on at the top to price fix the big hardware venders to make the cloud a more tempting option.

Speaking of the cloud, would you consider the cloud as a replacement to your physical servers? Or a hybrid if you have applications that need a local server?

 

[human7]

Speaking of the cloud, would you consider the cloud as a replacement to your physical servers? Or a hybrid if you have applications that need a local server?

Our company had a huge push to the cloud over the last few years. We have a ~10,000 CPU core on-prem cluster that was old and getting saturated, and we pushed a lot of our workloads to slightly smaller, temporary, spikey-workload clusters in the cloud.

And other teams within our company have done a lot of migrations, too. You can get a lot done, and the software and platforms are newer and better supported, so it isn't just about getting more or faster hardware.

But, now there's a lot of cost concerns, especially with the macroeconomic situations, and so some of the workload might be moved back on-prem. I've seen other articles about companies moving things back on-prem due to cost.

Frankly, the cloud providers are going to have increasing pressures to offer cheaper rates. Will they budge or just offer faster hardware at the same price and claim improved price performance? I dunno. Either way, there's a lot to be said for on-prem hardware that's been paid for. But there's a lot to be said about up-to-date and flexible software in the cloud, too.

 

[Burty117]

Our company had a huge push to the cloud over the last few years. We have a ~10,000 CPU core on-prem cluster that was old and getting saturated, and we pushed a lot of our workloads to slightly smaller, temporary, spikey-workload clusters in the cloud.

And other teams within our company have done a lot of migrations, too. You can get a lot done, and the software and platforms are newer and better supported, so it isn't just about getting more or faster hardware.

But, now there's a lot of cost concerns, especially with the macroeconomic situations, and so some of the workload might be moved back on-prem. I've seen other articles about companies moving things back on-prem due to cost.

Frankly, the cloud providers are going to have increasing pressures to offer cheaper rates. Will they budge or just offer faster hardware at the same price and claim improved price performance? I dunno. Either way, there's a lot to be said for on-prem hardware that's been paid for. But there's a lot to be said about up-to-date and flexible software in the cloud, too.

10,000 CPU core cluster! Damn! That's some big stuff, How big is your company? I work in the SME market so anything from a one man band to about 250ish employees and everything in-between. Moving to the cloud makes sense when you're really small as a M365 Premium License pretty much gives you everything you need.

Then once you start getting to the 150+ employees, being able to tie in AzureAD with WVD and Azure servers makes everything easy. The cost of hardware and software licensing has been a real turn off for this sized market. Plus lower requirement for an IT presence since no physical hardware needs looking after or updating regularly.

Not all our clients are like that though, I know of a few that have stuck to on-prem hardware because their Database requirements are crazy expensive in the cloud vs on-prem or they simply spent so much money on licensing VMware and Windows Datacenter they don't financially benefit going to the cloud at all.

 

[human7]

10,000 CPU core cluster! Damn! That's some big stuff, How big is your company? I work in the SME market so anything from a one man band to about 250ish employees and everything in-between. Moving to the cloud makes sense when you're really small as a M365 Premium License pretty much gives you everything you need.

Then once you start getting to the 150+ employees, being able to tie in AzureAD with WVD and Azure servers makes everything easy. The cost of hardware and software licensing has been a real turn off for this sized market. Plus lower requirement for an IT presence since no physical hardware needs looking after or updating regularly.

Not all our clients are like that though, I know of a few that have stuck to on-prem hardware because their Database requirements are crazy expensive in the cloud vs on-prem or they simply spent so much money on licensing VMware and Windows Datacenter they don't financially benefit going to the cloud at all.

Yeah it is a big cluster, but it is also easy to saturate. We are a mid sized company, and we have more than one cluster of a substantial size, but they are shared resources used by many different teams. There are times when our cluster is fully utilized and other times when rarely anything is on it (like on the weekends). It's this part which makes the cloud more attractive: you only have to pay for what you use, you don't have to over allocate to hit those peak times and then pay for it during the down times.

Even so, those charges quickly add up. It's the difference between a subscription model and a buy-it model: in the short term the subscription is always cheaper. In the long run, it usually isn't cheaper. But it's not just about cost, there are capabilities the cloud offers that we just don't have on-prem.

Our company isn't a startup, though, we've had datacenters for years. For a startup to build out that infrastructure, it would take a very long time for the cloud to be more expensive. And lots of other enterprises are moving back to a hybrid model. https://www.forbes.com/sites/peterb...o-on-premises-private-clouds/?sh=59fb946363cc