Apple's long-term demand for 3nm chips may be lower than expected

AppleInsider

Analyst Ming-Chi Kuo believes that weakness in Apple hardware demand through 2024 is causing a ripple effect that is reaching chip manufacturers, with orders for equipment falling by 30% in the next year.

15-inch MacBook Air

In a post on Medium on Wednesday, analyst Ming-Chi Kuo is examining extreme ultraviolet (EUV) equipment sales to chip manufacturers. Specifically, ASML may cut production of the expensive gear because Apple and other firms may not need as many chips as the industry forecast.

"In 2023, Apple's MacBook and iPad shipments declined significantly by approximately 30% and 22% to 17 million and 48 million units, respectively," Kuo writes. "The sharp decline is attributed to the end of work-from-home (WFH) demand and diminishing user appeal for the new specifications (Apple Silicon and Mini-LED)."

He goes on to say that there is a "lack of growth drivers" for MacBook and iPad.

This, paired with Qualcomm's lower demand for 3nm, plus lighter demand for Intel's 20A process and Samsung's 3GAP+, is a problem. The main DRAM manufacturers aren't expected to expand capacity until 2025 at the earliest and 2027 at the latest, depending on market conditions.

Apple has noted a contraction in the PC market due to higher-than-expected buys during the height of the COVID pandemic. It has weathered the storm better than competitors, but it remains unclear how long the demand will last.

Apple's M3 chip was originally rumored to arrive in April 2023, then fall 2023. More recent rumors peg the new Apple Silicon Mac chip based on a 3nm process to arrive in 2024, which is reasonable considering that Apple appears to have purchased all of TSMC's 3nm chip production through the end of 2024 for the iPhone.

Read on AppleInsider

Toortog

I think besides the economy the another part of the problem is chips have become so fast increases in performance are not noticeable by the average user.   Companies especially Apple kept creating background crap/apps to try and chew up some CPU cycle to create a need for more powerful chips, but the chips are handling it all just fine.    The only people really benefiting are people working in production level video.    Even people working in audio the faster chips are of little benefit.    I think this is why Apple is finally trying to get into the game market.  Hardcore gamers want and are willing to spend money on bleeding edge tech so Apple needs a presence in that market to sell its next generation chips. 

So selling more powerful chips are becoming harder sell. 

danox

Apple is doomed again? Despite the fact that Apple appears to have hit a home run with the iPhone 15 Pro, and 15 Pro Max, and the demand in China seems to be strong,
And you know Apple has hit a home run, when the comment sections are lit on fire with Samsung and Pixel users.

https://www.youtube.com/watch?v=q3iyXdebOvg

https://www.youtube.com/watch?v=vUn68iFMNXY

ApplePoor

For the majority of the customers not involved with high multi thread projects, their interest is in the single core speed of the M series because that is where they are operating most of the time. There is a less than 10% increase in the single core speeds going from M1 to M2. However, some of the same level M2 chips grabbed access to more memory at the same level.

So one wonders if TSMC can find another 10% increase in single core speed for the M3. An increase of 20% from M1 to M3 could possibly entice some folks to consider upgrading. However, we are rapidly approaching the point where, as humans, we are unable to perceive the speed increases. Is it happening now in a quarter of a blink of the eye versus half a blink of the eye?

PauloSeraa

Expectations are high for M3 because of moving to 3nm. If M3 performance doesn't meet with people's expectations (realistic or not), then sales will certainly be impacted.

M3 needs better GPU performance, Thunderbolt 5, and improved battery life all at the same time if its going to motivate anyone to upgrade from M1.

dewme

Can we get some second opinions on these forecasts and predictions? This is starting to seem like a Ming-Chi Kuo blog site. 

FileMakerFeller

I have serious doubts about this argument.

My gut feeling is that if the PC and mobile phone manufacturers aren't ordering as many 3nm chips there will be PLENTY of other device manufacturers happy to soak up the demand. I don't understand why Intel, Qualcomm and Samsung are purportedly not having success convincing their own customers to make the switch to the smaller-process chips, or selling to brand new customers. Per-unit costs are going to be higher, but the devices that can be built with the newer chips have so much to gain. Aren't we supposedly in the midst of the IoT revolution? Don't we have concerns from the US DoD that chip supply reliability is a strategic concern? What better time to secure manufacturing capacity than when commercial demand drops? Aren't the auto makers still demanding processing capability for their vehicles? Don't we have an AI boom that is processor-hungry? Digital currencies to mine? AAA games to play?

Frankly, if EUV equipment sales drop it's going to be because of the ban on selling that stuff to China and not because of reduced demand from the PC/phone sector affecting the chip manufacturers (which is also only an allegation at this stage). And even IF all of that is true it's likely to be a short-term dip as the world normalises now that covid has been mitigated.

CheeseFreeze

I started buying Apple hardware much less frequently over the years by asking the question: “do I really need this?”. Often, the answer is “no”, or “not really”.

The result is two 5+ year old iPads for my kids, the first gen iPad Pro being fine for my use, a base M1 Mac mini for my daily work, a 2018 MacBook Pro and two 6+ year iMacs. The only new/new/new device is the iPhone 15 Pro because my older phone was malfunctioning.

From the above list my disappointment is the iMacs not updating from Big Sur while their specs really should. It’s planned obsolesce. 

The iPads are for the kids and run fast enough; one of them is 64gb and that was a terrible choice since it runs out of space all the time. But still I’ll wait until it breaks before I switch. 

The iPad Pro is a bad product; expensive and doesn’t allow pro workflows. The normal iPad is a much better proposition. 

Then the MacBook Pro of them all is the one I would really want to replace, but most likely by a MacBook Air M3. I don’t need the power of a Pro device yet I like to benefit from hardware ray tracing and the overall speed increase of a third gen Apple silicon device. 

In the end most of us only need a new device when the old one is practically dying. The rest is just psychological manipulation by a brand 😅. 

avon b7

FileMakerFeller said:

I have serious doubts about this argument.

My gut feeling is that if the PC and mobile phone manufacturers aren't ordering as many 3nm chips there will be PLENTY of other device manufacturers happy to soak up the demand. I don't understand why Intel, Qualcomm and Samsung are purportedly not having success convincing their own customers to make the switch to the smaller-process chips, or selling to brand new customers. Per-unit costs are going to be higher, but the devices that can be built with the newer chips have so much to gain. Aren't we supposedly in the midst of the IoT revolution? Don't we have concerns from the US DoD that chip supply reliability is a strategic concern? What better time to secure manufacturing capacity than when commercial demand drops? Aren't the auto makers still demanding processing capability for their vehicles? Don't we have an AI boom that is processor-hungry? Digital currencies to mine? AAA games to play?

Frankly, if EUV equipment sales drop it's going to be because of the ban on selling that stuff to China and not because of reduced demand from the PC/phone sector affecting the chip manufacturers (which is also only an allegation at this stage). And even IF all of that is true it's likely to be a short-term dip as the world normalises now that covid has been mitigated.

Only a tiny fraction of the world's chip output is on cutting edge nodes. Of that production, the vast majority goes into phones. 

Nowadays, even phones don't really need cutting edge nodes either. At least not in the same way as a few years ago. Over 80% of the world's chip output is at 40nm or much higher (40nm-180nm and larger). When I looked at some numbers just a few years ago, around 98% was not on cutting edge nodes. 

At that time, TSMC had around 90% of that 2%.

With costs rising and demand slowing there is a risk that a negative catch 22 situation could arise where not enough revenues are being generated to fuel future, very expensive, development. ASML's latest lithography machines are rumored to cost $300 million. Twice as much as the previous generation. Then there is the age old problem of yield and process maturity (both in hardware and software). 

IoT, while simple, small and nimble, has to be cheap too and isn't financially viable on the latest nodes where available capacity is normally strained. 

In spite of all that, there is normally enough secondary demand on cutting edge nodes to keep that ship afloat. 

Apple and Huawei would typically be ramping orders for mass production to begin around April each year, for eventual release in September. Then early in the following year, Qualcomm, Nvidia etc would get a look in. 

The next earnings calls from ASML (EUV lithography orders are rumoured to be down), TSMC, Apple and Samsung will clear things up a lot.

The handset market specifically, has been a huge driver of cutting edge nodes. Laptops to a much lesser degree. That means any longer term downturn in that market will have an impact on the industry.

Also, Apple seems to be moving to a setup where they want to squeeze more life out of the previous years chips, leaving less demand for the latest and greatest (at least in the short term).