Short Course 1 from #iedm24 today. TSMC giving some detail about the state of the industry. Some highlights.
#1 3D Stacking is expected to go to 3 micron pitch by 2027.
#2 A16-based AI accelerators in 2027 with CoWoS-L and 12xHBM4. But even those 2025/2026 designs look cool - six compute die, 12xHBM3e. Next gen MI400 from AMD maybe?
#3 WiFi 8 to show life in 2027/2028. WiFi 8 at 10 Gbps may need as much as 30 TOPs and TSMC N4 to achieve peak perf. Could do it in N16, but big chip with lots of power.
Sucks that still so much WiFi 4 will still be being installed.
#4 Chiplet based smartphone SoCs are coming. DRAM goes on the side.
#5 The prolific use of interposers will move integrated voltage regulators to that interposer - both inductors and PMICs.
Intel kind of already does this with some generations of server CPUs - package on package designs.
#6 Big gains in the N2 era. GAAFETs rule
#7 In 2027, 50%+ of smartphones will have a >38 TOPs NPU on board - but less than 30% of notebooks.
#8 General server TAM will remain stagnant until 2027. AI Server TAM will grow 39% CAGR 22-27. Unit CAGR is 73%.
#9 Optical I/O / co-packaged optics are coming.
#10 HBM and CoWoS will come to automotive. Sounds somewhat insane, but here we are.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Ready for @intel #vision keynote with new CEO Lip-Bu Tan? Lots of eyes on this one. Starts on the hour. A 🧵
Some initial thoughts. LBT has to show he understands the company. He needs to showcase where Intel stands in key markets, and identify the direction to go in. If Pat was the aspirational, LBT has to be the pragmatist, and show it.
He's only been the role a month. I'm not expecting any major announcements or immediate changes in Intel's direction right now. He has some time for that for sure. LBT has to show how sharp he is, where he's focused, and what matters most.