Global Semiconductor Foundry Revenue Share: Q3 2023
Published Date: November 30, 2023
This page shows the quarterly revenue share for the top players in the global semiconductor foundry market from Q1 2022 to Q3 2023.
Global Foundry Market Share (%)
(*) Samsung includes foundry service for its internal logic IC business
This page provides a view on the global foundries revenue share from 2021 till 2023. Here are some highlights from Q3 2023:
TSMC recorded sequential growth in revenue in Q3 2023, helped by positive signs of recovery in the PC and smartphone markets, along with early signs of demand stabilization.
Automotive and industrial applications underwent inventory corrections in Q3 2023, and we expect the inventory correction period to linger into Q1 2024.
AI chip demand was robust in Q3 2023 and will remain strong in Q4 2023, evident from the AI GPU shortage. We believe this trend will extend into 2024.
Recovery in advanced nodes is expected to surpass mature nodes in terms of capacity utilization rate. Mature node foundries will face challenges from low utilization rates and increasing price competition going forward.
Read our foundry quarterly report for Q3 2023 here.
For detailed insights on the data, please reach out to us at sales(at)counterpointresearch.com. If you are a member of the press, please contact us at press(at)counterpointresearch.com for any media enquiries.
Weaker-than-expected macroeconomic situation continued to weigh on TSMC’s Q2 2023 business performance. Muted smartphone and PC/NB demand negatively impacted the overall utilization rate during the quarter. Though largely expected by the market, the company further cut its full–year revenue guidance on the weaker end demand expected for H2 2023. However, TSMC projects a strong AI demand in Q32023 and, going forward, sees itself as the key enabler for AI GPUs and ASICs that require a large diesize. We give ourtakes on the key points discussed during the earnings call:
Is AI semiconductor demand real?
Chairman(Mark Liu):Neither can we predict the near future, meaning next year, how the sudden demand will continue or will flatten out. However, our model is based on the data center structure. We assume a certain percentage of the data center processors are AI processors and based on that, we calculate the AI processor demand. And this model is yet to be fitted to the practical data later on. But in general, I think the trend of a big portion of data center processors will be AI processors is a sure thing. And will it cannibalize the data center processors? In the short term, when the capex of the cloud service providers is fixed, yes, it will. It is. But as for the long term, when their data service – when the cloud services have the generative AI service revenue, I think they will increase the capex. That should be consistent with the long-term AI processor demand. And I mean the capex will increase because of the generative AI services.
Adam Chang’sanalyst take: Supply chain checks reveal that cloud service providers such as Microsoft, Google, and Amazon aggressively invest in AI servers. NVIDIA is continuing to add orders for the A100 and H100 to the supply chain, echoing the strong momentum for AI demand. TSMC holds a significant market share in AI semiconductor wafer production, mitigating the risk of misjudging CoWoS capacity expansion concerning AI demand.
Akshara Bassi’s analyst take: Over the medium term, as hyperscalers continue to develop their own proprietary AI models and look to monetize through AI-as-a-Service and simiilar models, the infrastructure demand should remain robust.
Can AI semiconductor demand offset short-term macro weakness?
CEO (Che-Chia Wei):Three months ago, we were probably more optimistic, but now it’s not. Also, for example, China economy’s recovery is actually also weaker than we thought. And so, the end market demand actually did not grow as we expected. So put all together, even if we have a very good AI processor demand, it’s still not enough to offset all those kinds of macro impacts. So, now we expect the whole year will be -10% YoY.
Adam Chang’s analyst take: Although there is a lot of promise around AI, it would only account for around 6% of total revenues in 2023. Therefore, AI is not a panacea for broader short-term demand weakness.
Is TSMC CoWoScapacity enough to fulfill current AI demand?
CEO (Che-Chia Wei): For AI, right now, we see very strong demand, yes. For the front-end part, we don’t have any problem to support, but for the back end, the advanced packaging side, especially for the CoWoS, we do have some very tight capacity to — very hard to fulfill 100% of what customers needed. So, we are working with customers for the short term to help them to fulfill the demand, but we are increasing our capacity as quickly as possible. And we expect these tightening will be released next year, probably toward the end of next year. Roughly probably 2x of the capacity will be added.
Adam Chang’s analyst take: Due to TSMC’s CoWoS capacity constraints, the company is finding it challenging to fulfill the strong AI demand from customers,, including NVIDIA, Broadcom, and Xilinx, at the moment. NVIDIA is actively seeking second- source suppliers as TSMC looks to outsource some of its production.
CEO (Che-Chia Wei): N3 is already involved in production with good yield. We are seeing robust demand for N3 and we expect a strong ramp in the second half of this year, supported by both HPC and smartphone applications. N3 is expected to continue to contribute mid-single-digit percentage of our total wafer revenue in 2023. Our N2 technology development is progressing well and is on track for volume production in 2025. Our N2 will adopt a narrow sheet transistor structure to provide our customers with the best performance, cost, and technology maturity.
Adam Chang’s analyst take: Apple is the sole customer expected to adopt TSMC’s 3nm technology in its A17 Bionic and M3 chips during 2023. The Qualcomm Snapdragon 8 Gen 4 processor is also anticipated to join the TSMC 3nm family (N3E) in 2024. Moreover, Intel is likely to adopt TSMC’s 3nm technology for its Arrow Lake CPU, scheduled to launch in H2 2024.
Q2 2023 results beat slightly: TSMC reported $15.67 billion in sales, slightly above the midpoint of guidance. EPS beat consensus due to higher non-operating income. Both GPM and OPM slightly beat guidance thanks to favorable FX and cost control efforts.
Q3 2023 guidance in line: The management guided $16.7-$5 billion (+9% QoQ at midpoint), gross margin in the range of 51.5%-53.5%, and operating margin in the range of 38%-40%. The gross margin dilution resulting from the N3 ramp-up would be 2-3 percentage points in Q3 2023 and 3-4 percentage points in Q4 2023. This impact would persist throughout the entire year of 2024, affecting the overall gross margin by 3-4 percentage points. Notably, this dilution is higher than the 2-3 percentage points gross margin dilution experienced during the N5’s second year of mass production in 2021.
2023 revenue guidance revised down but expected: TSMC revised down the full-year revenue guidance to -10% YoY. The management sees weaker-than-expected macroeconomics in H2 2023 affecting the demand for all applications except for AI.
Strong AI demand, 50% revenue CAGR forecast: AI revenue currently makes up 6% of TSMC’s total revenue. The company anticipates a remarkable compound annual growth rate (CAGR) of nearly 50% from 2022 to 2027 in the AI sector. As a result of this significant growth, the AI revenue percentage share in TSMC’s total revenue is projected to reach the low teens by 2027.
CoWoS capacity expected to double by 2024 end: TSMC is experiencing strong demand in the AI sector, with sufficient capacity for the front-end part but facing challenges in advanced packaging, particularly CoWoS.It is working with customers to meet demand in the short term while rapidly increasing capacity which it expects to double by the end of 2024, easing the current tightness.
Qualcomm takes the top spot in terms of cost contribution, accounting for over 34% of the model’s BoM cost.
Qualcomm and Samsung combined contribute more than 65% of the component cost in the Galaxy S23 Ultra.
Featuring Qualcomm’s custom Snapdragon chipset, Samsung’s Galaxy S23 Ultra makes a considerable leap in computing performance with its shift to TSMC’s 4nm process node.
Producing an 8GB+256GB Galaxy S23 Ultra (Sub-6GHz) variant costs Samsung around $469, according to the latest bill of materials (BoM) analysis by Counterpoint’s component research service. The major components driving cost in the smartphone are the SoC, display and camera subsystem. Due to excess inventory and supply, components related to the RF sub-system and memory were subjected to a cost decrease.
Qualcomm and Samsung’s design wins
The S23 Ultra further builds upon Qualcomm’s design, showcasing a customized version of the Snapdragon 8 Gen 2 chipset, manufactured on TSMC’s 4nm process node. Samsung has chosen Qualcomm chipsets due to enhanced cellular support, increased performance gain from both the CPU and GPU, and better battery life. The GPU also has support for raytracing and has gained a slight uplift of 39MHz clock speed.
Qualcomm’s share in the S23 Ultra has increased to an all-time high after attaining design wins for the fingerprint sensor IC, key power management ICs, audio codec, RF power amplifiers, Wi-Fi + Bluetooth, GPS and Sub-6GHz transceiver.
Samsung is the second largest beneficiary. It is an exclusive supplier of the 256GB NAND flash and the 6.8-inch AMOLED display for the S23 Ultra. The display can sustain 1750 nits of peak brightness and has a resolution of 1440 x 3088 pixels that allows the users to view pictures and videos in sharp detail. The 120Hz LTPO panel also supports adaptive refresh rate.
In the camera sub-system, the design wins are shared between Samsung (SEMCO) and Sony. Samsung provides the 200MP wide-angle camera (S5KHP2) and the 12MP selfie camera (S5K3LU), while Sony offers the 12MP Ultrawide (IMX564), 10MP Telephoto and Periscope Telephoto (IMX754) sensors.
Other component suppliers
Silicon Mitus and Maxim are the providers of power management ICs that support the regulation of power for display and other key components.
For sensing components, STM has registered design wins related to the laser autofocus module, accelerometer, gyroscope, barometer, and touch panel controller. The battery is packaged by Samsung and the cell is provided by ATL. The quick charging IC, which charges up to 45W, is sourced from NXP while the 15W wireless charging IC is from Convenient Power.
Samsung’s sourcing strategy and choice of components are enabling the brand to have a competitive edge in terms of cost efficiency.
The legislation includes over $52 billion in subsidies for semiconductor manufacturing.
So far, investments worth over $200 billion have been announced for manufacturing.
New foundries from TSMC, Intel, Samsung, Micron and Texas Instruments are underway.
Last August, the US Congress passed the CHIPS Act, the largest piece of industrial policy signed into law in the US in a generation. The legislation includes over $52 billion in subsidies for semiconductor manufacturing. Over the past two decades, the share of semiconductors manufactured in the US has steadily fallen while advanced semiconductor manufacturing has become increasingly concentrated in just a handful of cities, primarily in Taiwan and South Korea.
The outbreak of COVID-19 wreaked havoc on supply chains as factories shuttered, while the fallout of an unexpected winter storm in Texas further disrupted semiconductor manufacturing. Compounded by an explosion of demand as companies, schools and government offices pivoted to work from home, the chip shortage set in, shaving a percentage off of GDP growth according to the White House. In summary, the events of the past few years have made abundantly clear the foundational role that semiconductors play in today’s economy, and the costs that accompany constrained supply.
All this triggered the finalization and passage of the CHIPS Act. The US government was convinced that the country’s supply of semiconductors, which power everything from washing machines, smartphones and cars to supercomputers and hypersonic missiles, faced an unacceptable bottleneck at a pivotal moment. The US is almost entirely dependent on Taiwan for the production of advanced chips even as cross-strait tensions reach new highs and US-China relations new lows. Indeed, concerns over a potential conflict in the South China Sea as well as between North Korea and South Korea, not to mention acts of God, impacting chip supplies convinced the US government about boosting the country’s semiconductor manufacturing. But with higher labor costs and plenty of red tape, semiconductor manufacturers needed added incentives to make the transition worthwhile. The CHIPS Act has done just that, sparking a wave of private investment.
Since the CHIPS Act was passed, investments worth over $200 billion have been announced for manufacturing capacity in the US, with new foundries underway from TSMC, Intel, Samsung, Micron and Texas Instruments. But while the CHIPS Act has lit the flame under private companies to act, some of the legislation’s provisions and shortcomings could lead to its undoing. These projects are faced with red tape and regulations that will cause them to linger before coming on line. Besides, while funding has been made available for workforce education and job training, the scale of the need for new employees and the lack of the required workers and skills will likely pain semiconductor firms for years to come. One solution would be to raise the number of visas available for skilled workers from abroad. Another would be to provide targeted assistance to students pursuing degrees in related fields. Finally, the Act itself fails to address the reason why semiconductor manufacturing left the US in the first place – American labor costs and regulations make production in the US more expensive than elsewhere. Once the funding runs dry, how will American semiconductor manufacturing remain competitive? Additional burdens on employers to guarantee union wages and provide child care certainly won’t make these projects anymore cost-competitive.
While flawed, the CHIPS Act is a major stepping stone to creating secure, resilient supply chains that will insulate the country from many outside shocks. This is a step in the right direction, but more must be done if the country wants to win the semiconductor manufacturing marathon and avoid fizzling out after the starting sprint.
For more information about the CHIPS Act, a detailed report on the legislation can be found here.
• TSMC has joined other semiconductor companies in giving conservative guidance due to the expected industry downturn in the next few quarters. This includes declining the capacity utilization rate from Q4 2022, capex cuts in 2022 and 7/6nm order adjustment.
• TSMC’s 7/6nm contributed 26% to the total revenue in Q3 2022. Based on our analysis, smartphone and HPC-Computing (PC and server CPUs, and GPUs) are the two largest segments in this technology node, accounting for 32% and 38%, respectively, of the total wafer shipment volume for 2022.
• TSMC has blamed the 7/6nm softness on cyclical inventory adjustment and product delays from smartphone and PC clients. It looks the inventory cycle will persist into 2023, mainly from AP/SoCs for mainstream 5G smartphones.
• We agree with TSMC’s view on positive drivers of new product migration on the 7/6nm node, like Wi-Fi, RF and SSD controller ICs, after the inventory cycle in 2023.
Although TSMC is still confident about its business growth in 2023, the company admitted during its investor conference for Q3 2022 that inventory headwinds would hit its near-term sales outlook with declining utilization rates in certain geometry nodes. The full-year capital expense revision for 2022 from $40 billion to $36 billion echoed its conservative views on delaying new capacity builds amid the global semiconductor downturn, including the withdrawal of the new 7/6nm line in Fab 22 due to uncertain market demand.
While TSMC expected solid demand for its current 5/4nm nodes with a high utilization rate in Q4 2022 too, it gave conservative guidance for 7/6nm on projections of a declining utilization rate in the next few quarters. TSMC attributed this to smartphone weakness, as well as product delays in PC-related chipsets. The company expects the inventory adjustment cycle for all TSMC technology nodes and chip productions to likely persist into 2023.
TSMC’s 7/6nm node was its largest revenue component before Q3 2022, contributing to nearly 30% of its business during the first three quarters of 2022. TSMC owned a total of 145-150 KWPM (thousands of wafers per month) installed capacity on 7/6nm during H1 2022 but will adjust the production plan for some equipment tools in H2 2022, partially taking into account the deteriorating market demand for smartphones and PCs.
By breaking down TSMC’s wafer shipments for 7/6nm products, we found HPC-related products (including PC and server CPUs, discrete GPU, data center accelerator and ASIC/FPGA) accounted for 38% in 2022, followed by smartphone-related (mainly on AP/SoCs) chipsets at 32%. MediaTek, AMD and Qualcomm appeared to be the top three clients in this category.
Supply chain inventory to remain high at 2022 end
Reviewing the inventory level for smartphone AP/SoCs, which is the major application of advanced foundry nodes (10nm and below), we believe the order corrections to chipset vendors will be under greater pressure from H2 2022 amid weakening end-market (sell-through) data point. The chip production lead time is as long as over four months in advanced nodes, leading to a lagged effect of wafer output reductions earliest from Q4 2022 in smartphone AP/SoCs, CPU/GPU and AI processors. Accordingly, the chipset-level inventory cycle appears to have just started from H2 2022 and will persist into the remainder of the year or into the first half of 2023, as TSMC commented during the investor conference call.
More of a cyclical adjustment issue in 7/6nm
As the global foundry industry’s utilization rate has reached its peak level in mid-2022, the downtick will bring down business in all aspects in the next few quarters before any signs of improvement emerge in inventory levels across the semi supply chain. All of TSMC’s technology nodes will be inevitably impacted, particularly 7/6nm with its higher concentration of smartphones and consumer PCs. The utilization rate here will drop to 80%-90% in the next two to three quarters prior to the demand recovery driven by mainstream 5G smartphone AP/SoCs and Intel’s acceleration of its Meteor Lake CPU tile orders on TSMC. Besides, we agree with TSMC’s view on positive drivers of new product migration on the 7/6nm node, such as Wi-Fi, RF and SSD controller ICs, after the inventory cycle in 2023.
TSMC gained 200 bps in terms of market share in Q2 2022, driven by leading technology nodes for iPhone AP and HPC-related ICs. The capacity utilization rate in Q2 in leading nodes was still at 95%-100% at TSMC. Samsung lost 100-200 bps during the quarter due to weaker demand for Qualcomm’s flagship smartphone SoCs. UMC, GF and SMIC managed to maintain their shares with high capacity utilization rates.
Foundry Revenue Share by Technology Nodes
5/4nm became the second largest node in Q2 2022 in terms of revenue share, accounting for 15% of industry sales, which were driven by smartphone SoCs (Apple, Qualcomm and MediaTek), ARM-based CPUs (Apple) and newly launched GPU accelerators from NVIDIA. 7/6nm remained the largest node, with increasing applications from CPU, GPU and networking processors. Sales from matured nodes started slowing down in Q2 2022 due to inventory adjustment from driver ICs and CMOS sensors.
Use the button below to download the high resolution PDF of the infographic:
COVID-19, along with the rise of advanced capabilities such as 5G, AI and imaging technologies, has catalyzed the semiconductor demand for the last over two years. Monitoring the contribution of upstream players in the semiconductor value chain, which are actually building technologies and capacities, has become extremely important.
TSMC is a great benchmark for the health of the semiconductor industry considering it manufactures 70% of all key smartphone chipsets. The company posted record earnings in Q2 2022 with growing advanced semiconductor content in processing (AI, GPU, SoC) and connectivity (5G) being the key factors.
Key financial highlights:
Net revenue increased 37% YoY to $18.2 billion driven by high-performance computing (HPC), IoT and automotive-related demand.
Gross margin and operating margin were at 59.1% and 49.1% respectively, up 3.5 percentage points on a favorable foreign exchange rate, cost improvement and value selling.
From the geographical perspective, North America accounted for the highest share (64%) of total net revenue.
TSMC wafer revenues share
Smartphones and HPC represented 38% and 43% of net revenues respectively, while IoT, Automotive, Digital Consumer Electronics (DCE) and Others represented 8%, 5%, 3% and 3% respectively.
HPC surpassed Smartphones in revenues thanks to Nvidia, Intel, AMD and others.
TSMC’s reliance on Apple, Qualcomm and Mediatek was lesser as HPC surpassed Smartphones in revenue contribution.
Automotive semiconductor content was the dark horse.
5nm process technology contributed 21% of total wafer revenues in Q2 2022 while 7nm accounted for 30%.
Combined revenue from advanced process nodes with 5nm and 7nm accounted for 51% of total wafer revenues, thanks to growing capex, making it very difficult for current and potential competition to catch up at least in next 10 years.
Double-digit growth was seen in matured nodes thanks to the rising need for chipsets in the IoT and automotive.
N2 and N3 updates
N2 node will implement the platform scaling concept wherein benefits of power delivery schemes, advanced packaging and chiplet will be utilized to control cost and have an overall advantage.
N3 node will be the longest node to be used before migrating to N2 due to the introduction of TSMC FINFLEX architectural innovation, which offers flexibility to customers to create designs precisely tuned for their needs with functional blocks implementing the best-optimized fin configuration and integrated into the same chip.
The introduction of 3nm nodes will begin in H2 2022 and adoption by customers and revenue contribution will start in Q1 2023. The introduction of 3nm nodes will lower the gross margin by 2%-3% in 2023.
While the capex is growing, some of it will be spread out over quarters with the WFE vendors struggling with backlogs as building fab equipment also requires semiconductors! This will help TSMC realize healthy margins for the coming quarters and offset any gross margin decline due to N3 introductions.
TSMC’s net revenue will cross $75 billion in 2022, which means it will surpass Intel’s revenues.
HPC will drive TSMC’s revenue growth in the long term and achieve a 15%-20% CAGR.
3D IC design solution System on Integrated Chips (SoIC) will account for a significant share of revenue in the long term due to its extensive application in HPC.
Efforts to resolve tool delivery schedule challenges in advanced and mature nodes through discussions with entire supply chain partners remains a top priority.
Growing silicon content, shipments and ASP will drive revenue growth in the long term.
Inventory adjustment will continue till Q1 2023 and ease off by H2 2023. However, long-term semiconductor demand will be firm.
TSMC reported stronger Q1 2022 revenue driven by HPC (clients including Apple, AMD, Nvidia) which exceeded smartphone as its largest business segment for the first time of company history. Also the upside results came from UMC and SMIC, both delivered higher sales from increasing wafer price and local client demand. Relatively, the market share on Samsung and GF were stable, due to the constrains of capacities and lower production yields from some products in the advnaced nodes.
Foundry Revenue Share by Technology Nodes
The largest technology node (in revenue) was 7/6 nanometer during Q1 2022, accounting for 18% of the total foundry industry TAM. The main products in 7/6nm include smartphone AP/SoC, tablet APUs, GPUs, and CPUs that TSMC
dominated the market. The node of 16/14/12nm is the second largest node, primiarily for smartphone RF IC / 4G SoCs, wearable processors, SSD controller, and PC-related ICs. TSMC, Samsung and GlobalFoundries are the major vendors in 1xnm.
Use the button below to download the high resolution PDF of the infographic:
TSMC reported its Q1 2022 revenue at $17.6 billion (up 12% QoQ). Among its different segments, the high-performance computing (HPC) segment posted the highest growth rate at 26% QoQ.
HPC accounted for 41% of TSMC’s Q1 2022 revenue, while the smartphone segment accounted for 40%. It is the first time in TSMC’s history that HPC has become the largest sales contributor during one fiscal quarter.
TSMC’s major HPC customers in 2021 likely included AMD (including Xilinx), Nvidia, Broadcom, Marvell and Intel besides Apple as the top-selling one due to its Apple Silicon (M series) in PCs and tablets.
Intel appeared to be the next major HPC customer by outsourcing more graphic silicon (discrete Arc series and iGPU tiles) and other CPU cores this year. Accordingly, it would be the second-largest customer in TSMC’s 3-nanometer node in 2023.
TSMC’s HPC segment is expected to account for 42% of its total revenue in 2022, including wafer and 2.5D/3D packaging services.
Before TSMC’s investor conference on April 14, semiconductor industry experts were waiting for the company to tune down its annual growth outlook due to weakening consumer sentiment on consumer electronic products. Counterpoint projects the global smartphone sell-through units to decline by over 10% YoY in Q1 2022, impacted by a slowing replacement cycle, geopolitical uncertainties, COVID-19 lockdowns in China and inflation concerns. Recent data on smartphone order cuts implies the risk for foundry wafer demand in the next few quarters and the previous IC shortage concerns have been improved.
Surprisingly, TSMC beat market expectations not only in its Q1 2022 financial results (with gross margin improving to 56%), but the upside of full-year sales guidance also approached 30% YoY growth. While the demand softness from smartphones has been known, TSMC believes the momentum of HPC and automotive businesses is getting stronger in 2022 and will offset the uncertainties from consumer electronics. Notably, TSMC reported $7.2-billion HPC sales in Q1 2022, exceeding the sales from smartphone clients to become the largest business of the company for the first time in history.
Other than retaining its revenue outlook for 2022, TSMC reiterated its capital spending budget at $40-$45 billion during the year, with a focus on leading-edge technology nodes (5 nm and below). The longer lead time of equipment/tool production becomes more challenging in the light of labour crunch and IC component shortages (like FPGA) from global suppliers. But TSMC has demonstrated its solid execution in ramping up new fabs as the capacity expansion plans are largely unchanged.
HPC silicon demand is on a multi-year, structural growth trajectory
The silicon demand of HPC-related ICs is well above that of the end devices and equipment. In 2022, global PC shipments are likely to decline 0%-5% YoY. The server shipments growth rate is also expected to be modest (5%-8% YoY). At the same time, 5G infrastructure growth is not accelerating in many countries, with the growth cycles of edge computing or open RAN being at the initial stages.
So, why TSMC and its HPC customers are expected to deliver a very robust growth outlook in the next 1-2 years? The answer can be found in the mega trends of digital transformation since COVID-19, which have witnessed a boost in capital spending mainly by US-based cloud service providers (CSPs) expanding their global data centre footprints. The total capex by global CSPs, based on Counterpoint estimates, may increase 23% YoY in 2022. In the three years to 2025, the annual compound growth rate (CAGR) will be in double digits. The acceleration in mega data centre buildouts is for AI services, Metaverse and autonomous driving in need of the most advanced semiconductor components. The AI accelerator IC market, which has been dominated by Nvidia, is a typical example of the market where the demand CAGR is likely to exceed 30% in the next few years as GPU, CPU, FPGA and ASIC compete for AI training and interference applications.
All main processors in server and data centres today apply foundry nodes below 10 nm. As computational performance improves significantly and demands more transistors, the roadmap of node migrations beyond 10 nm will get accelerated in the HPC semiconductor industry. In 2022, both AMD and Nvidia’s new graphic processors will move to 4 nm and 5 nm nodes. Further, we expect the next phase of 3 nm generation processors to start to ramp up after 12-18 months. Besides, the increases in larger die size in processors implies more wafer demand and higher dollar content in advanced packaging technologies that TSMC offers for complete solutions.
In our view, the HPC semi demand in foundry remains the seller’s market in the next few years, as TSMC captures over 90% of market opportunities, including Intel’s CPU outsourcing orders. Competition in the HPC semi market from the global top three chipmakers – Intel, AMD and Nvidia – is intensifying by entering new areas, like Intel’s entry into the discrete GPU market and Nvidia’s plan for ARM-based server CPUs, leading to strong wafer demand in advanced nodes.
TSMC offers the leading-edge technology to enable these HPC applications to pursue the best PPAC (performance, power, area and cost) over each technology geometry. This has become TSMC’s fastest-growing business segment in the past three years from the revenue perspective. In Q1 2022, it reported HPC business sales reaching $7.2 billion with nearly 60% YoY growth, overtaking smartphones as the company’s largest segment. While TSMC did not disclose any client information, we believe AMD, Nvidia, Broadcom, Marvell and Intel were the major traditional HPC customers during the quarter, with Apple as the top-selling one due to its Apple Silicon (M series) in PCs and tablets.
5/4 nm is the mainstream today, 3 nm to ramp up from 2023
The reason why TSMC’s revenue is increasing from the HPC segment is that the majority of wafers in HPC applications apply the leading-edge nodes (from 7 nm and 6 nm to 5 nm and 4 nm in 2022). Wafer prices at these nodes are multiple times higher than the foundry industry ASPs due to more expensive equipment tools. During Q1 2022, the mainstream of HPC ICs was at 7/6-nm, including discrete GPUs, data centre GPUs (AI accelerators), x86 CPUs, ARM CPUs for client devices (tablets and laptops) and specific processors like crypto mining ASICs. TSMC will also start to ramp up multiple new CPUs/GPUs at 5/4 nm in H1 2022, leading to its robust wafer demand likely through the end of the year.
Based on our estimates, in 5/4-nm nodes, the total wafer demand from HPC products (see Exhibit 2) will account for 37% of the total foundry (excluding IDM) industry capacity in 2022. Smartphone AP/SoCs will remain the largest portion of wafer demand (over 50%) in this node during the year. With new product cycles from Nvidia and AMD in H2 2022, more wafers will be procured into 2023 as HPC might reach half of the total wafer capacity at 5/4 nm in the next year.
During its Q1 2022 financial result conference, TSMC reiterated 3 nm would enter mass production in H2 2022, with wafer sales contribution from the beginning of 2023. As we all know that the new iPhone application processor (A16) will adopt 4 nm instead of 3 nm due to initial supply constraints from TSMC, HPC products will be the main body of 3 nm in H1 2023, especially the demand from Intel’s wafer outsourcing in its new CPU line-ups. We believe Intel would be the second-largest client after Apple on TSMC’s 3 nm in 2023 as the volume of outsourced wafer orders is highly subject to Intel’s in-house EUV availability in the early stage. HPC applications will be a stable part (Exhibit 3) in 3 nm with 40%-60% of total foundry wafer demand during 2023-2025.
The crossover point between HPC and smartphone sales at TSMC, observed in Q1 2022, is more like the beginning of the future trend where semiconductors will shift growth drivers from mobile devices to multiple computing focus areas for the next wave of technology applications. We estimate TSMC’s HPC segment to account for 42% of its total revenue in 2022, including wafer and 2.5D/3D packaging services.
TSMC has announced a capital spending of $40-44 billion for 2022 keeping in view its massive new fab construction plans. This is above the industry forecast of $38-42 billion.
According to TSMC, 70-80% of capex in 2022 will be used for advanced node technologies (7nm and below), 10-20% for specialty technologies and 10% for advanced packaging.
If we factor in TSMC’s higher spending as well as greater sales growth (25-29% YoY) in 2022, we see the WFE capital intensity (capex to sales) for logic (non-memory) semiconductor reaching 23% in 2022-2023, the highest in the past 20 years.
TSMC’s massive capacity plan also implies higher wafer demand in advanced nodes, largely used in smartphone processors today. But we expect the demand from high performance computing (HPC) to exceed that of smartphones from 2023 at TSMC.
Before TSMC’s investor conference on January 13, equipment vendors expected TSMC to raise its capital spending in 2022 based on the company’s multiple investment plan announced in the second half of 2021. TSMC reported $30 billion of capital spending in 2021, about two times larger than Intel’s from the WFE (front-end wafer equipment) perspective. Based on the new guidance, it would spend $40-44 billion in 2022, which is above the industry forecasts.
In our view, TSMC’s higher capital spending implies the company’s more aggressive capacity build-up plan for (1) 3nm and 2nm nodes, as they might see greater demand from Apple and Intel beyond 2023, in addition to the CPU/GPU growth from data centers (2) 3D packaging, due to more demand from HPC clients (3) Specialty nodes, especially at 28/22nm.
Exhibit 1: TSMC Revenues vs CAPEX Trends 2016-2021
TSMC’s major capital spending items in 2022 include:
P5-P8 in Fab 18 (Tainan, Taiwan) ramp-up, mainly for 4/5nm and 3nm
Fab 20 (Hsintsu, Taiwan) ground-breaking, built initially for 2nm production
Kumamoto (Japan) Fab ground-breaking, primarily for local clients at 28/22nm
Kaoshiung (Taiwan) Fab ground-breaking, mainly for 7nm in the initial stage
Exhibit 2: Capital Intensity Ratio for Logic Semiconductor (Non-Memory) ,, 2010-2024
Implications for global logic semiconductor industry’s capital intensity ratio
Overall, the foundry industry is expected to account for 30-35% of the global logic (non-memory) semiconductor wafer production in 2022-2023. TSMC leads here by capturing two-thirds of the foundry market. Therefore, TSMC’s higher spending outlook indicates greater semi industry capex as well in 2022, and likely in 2023, if we also consider moderate expansions from other foundry/IDM players.
We usually use the capital intensity ratio (capex divided by sales) to track the semiconductor industry cycle to find early indications of future market growth. The chart above has two lines – the blue one for the entire logic semi industry (foundry and IDM) and the red one for TSMC and Samsung Foundry, the two major players dominating the capacity in leading-edge nodes. Factoring in TSMC’s new forecast for 2022 and industry outlook from increases in other new fab constructions, the capital intensity ratio would reach 23% in both 2022 and 2023, the highest in the past 20 years.
What are the projections for the semiconductor industry in 2022-2023?
Since the middle of last year, the supply chain has started to reflect concerns about the oversupply of fab capacities going into 2022, if we only look at the normal product cycles for consumer electronic devices such as smartphones and PCs. TSMC’s big announcement clearly indicates there are more structural growth drivers in semiconductors beyond 2021. Its clients too seem to be more optimistic about the silicon wafer demand amid the mega trend of digital transformation. If we use the capital intensity ratio to forecast the production cycle, it appears we won’t reach the peak before 2024 when most of the new capacities (matured and advanced nodes) start mass production. The current high inventory may only suggest a near-term correction in orders for certain components. 5G applications, HPC and automotive all require more silicon content going forward and will offset the saturated demand in smartphones.
In order to access
Counterpoint Technology Market Research Limited (Company or We hereafter) Web sites, you may be asked to complete a registration form. You are required to provide contact information which is used to enhance the user experience and determine whether you are a paid subscriber or not.
When you register on we ask you for personal information. We use this information to provide you with the best advice and highest-quality service as well as with offers that we think are relevant to you. We may also contact you regarding a Web site problem or other customer service-related issues. We do not sell, share or rent personal information about you collected on Company Web sites.
How to unsubscribe and Termination
You may request to terminate your account or unsubscribe to any email subscriptions or mailing lists at any time.
In accessing and using this Website, User agrees to comply with all applicable laws and agrees not to take any action that would compromise the security or viability of this Website. The Company may terminate User’s access to this Website at any time for any reason. The terms hereunder regarding Accuracy of Information and Third Party Rights shall survive termination.
Website Content and Copyright
– Passwords are for user’s individual use
– Passwords may not be shared with others
– Users may not store documents in shared folders.
– Users may not redistribute documents to non-users unless otherwise stated in their contract terms.
Changes or Updates to the Website
Accuracy of Information:
While the information contained on this Website has been obtained from sources believed to be reliable, We disclaims all warranties as to the accuracy, completeness or adequacy of such information. User assumes sole responsibility for the use it makes of this Website to achieve his/her intended results.
Third Party Links:
This Website may contain links to other third party websites, which are provided as additional resources for the convenience of Users. We do not endorse, sponsor or accept any responsibility for these third party websites, User agrees to direct any concerns relating to these third party websites to the relevant website administrator.
Cookies and Tracking
We may monitor how you use our Web sites. It is used solely for purposes of enabling us to provide you with a personalized Web site experience.
This data may also be used in the aggregate, to identify appropriate product offerings and subscription plans. Cookies may be set in order to identify you and determine your access privileges. Cookies are simply identifiers. You have the ability to delete cookie files from your hard disk drive.