In a report published by TrendForcewe receive our second confirmation that serial production of Intel’s tGPU for 14th generation Meteor Lake CPUs has started.
Intel’s 3nm tGPU Mass Production Lag Hurts TSMC, 14th Gen Meteor Lake Delayed To Late 2023?
Earlier it was rumored that Intel could delay its 14th Gen Meteor Lake CPUs to late 2023, and it looks like we’re getting similar reports from Trendforce sources close to Taiwanese chipmaker TSMC. It is reported that Intel originally planned to start mass production of the tGPU (Tiled GPU) on the 14th generation Meteor Lake CPUs from the second half of 2022. This was later pushed back to the first half of 2023 due to product design and process review issues, but now we may be looking at an even further delay.
According to TrendForce research, Intel plans to outsource the tGPU chipset in Meteor Lake to TSMC for manufacturing. Mass production of this product was originally planned for 2H22, but was later pushed back to 1H23 due to product design and process verification issues. Recently, for some reason, the product’s mass production schedule was pushed back again to late 2023, almost completely scrapping the 3nm production capacity originally booked for 2023, leaving only a marginal amount of wafer input for technical review.
However, the development status of Intel’s own Intel 4 process and the associated outsourcing situation are still important potential growth drivers for TSMC. If Intel 4 doesn’t make mass production as planned, Intel could outsource its compute tiles to TSMC, which will greatly fuel growth in 2024. However, if the Intel process develops smoothly, the possibility remains that the company will manufacture related products itself and cancel orders from TSMC.
According to the latest data, TSMC may not start mass production on Intel’s 3nm tGPU before the end of 2023. A report from Digitimes last month had something similar to say. it was said that Intel’s CEO, Pat Gelsinger, would do so visit TSMC yourself to revise its outsourcing plan with the Taiwanese chipmaker. Since Intel was a big 3nm customer, a delay would potentially hurt and slow down TSMC. Although there are other major suppliers for 3nm such as Apple, AMD and Qualcomm, their products are not expected to start production before 2024.
It also highlights that Intel’s own “Intel 4” processing node might have some issues, and if it couldn’t produce viable volume for Meteor Lake CPUs, Chipzilla might reconsider offloading other IPs like Compte Tile to TSMC. That could help TSMC continue to grow revenue, but based on Intel’s recent claims, it looks like progress is being made on their internal process node and showing good results, but we’ll see when we start 2023.
With Raptor Lake in the second half of this year and Meteor Lake in 2023, we’re building on Alder Lake’s leadership and illustrating how our innovative design choices can drive leadership performance even before we restore industry-leading transistor technology.
And in 2023, we will deliver our first disaggregated CPU based on Intel 4, Meteor Lake, which is in good shape both in our labs and in our customers’ labs.
Yes. Maybe if we quickly go through the five nodes, Intel 7 ready, volume shipments, we said five nodes, four years, Intel 7, 35 million units, you can tear one apart. I’m sure our competitors have done teardowns on it, done. Intel 4, right, we said is, hey, Meteor Lake, looks good at this point.
Pat Gelsinger, Intel CEO (Q2 2022 Earnings Call)
The tGPU will be one of the main components of the 14th generation Meteor Lake CPUs, with Arc graphics architecture to bring Intel on par with AMD’s and Apple’s on-die GPU solutions. This is all just a rumor for now though Recently leaked platform details for Meteor Lake CPUs confirm that the consumer launch is expected sometime in the second half of 2023.
Intel Mobility CPU Lineup:
CPU family | meteor sea | Raptor Lake | Erlensee |
---|---|---|---|
process node | Intel 4 “7nm EUV” | Intel 7 ’10nm ESF’ | Intel 7 ’10nm ESF’ |
CPU architecture | Hybrid (triple core) | Hybrid (dual core) | Hybrid (dual core) |
P-Core architecture | Redwood Bay | Raptor Bay | Golden Bay |
E-Core architecture | crestmont | Gracemont | Gracemont |
top configuration | 6+8 (H series) | 6+8 (H series) | 6+8 (H series) |
Max cores/threads | 14/20 | 14/20 | 14/20 |
Planned lineup | H/P/U series | H/P/U series | H/P/U series |
GPU architecture | Xe2 Battle Mage ‘Xe-LPG’ | Iris Xe (Gen 12) | Iris Xe (Gen 12) |
GPU Execution Units | 128 EUs (1024 colors) | 96 EUs (768 colors) | 96 EUs (768 colors) |
memory support | DDR5-5600 LPDDR5-7400 LPDDR5X – 7400+ |
DDR5-5200 LPDDR5-5200 LPDDR5-6400 |
DDR5-4800 LPDDR5-5200 LPDDR5X-4267 |
Storage capacity (max.) | 96GB | 64GB | 64GB |
Thunderbolt 4 ports | 4 | 2 | 2 |
WiFi capability | WiFi 6E | WiFi 6E | WiFi 6E |
TDP | 15-45W | 15-45W | 15-45W |
begin | 2H 2023 | 1H 2023 | 1H 2022 |