Intel has significantly amped up its roadmap since Pat Gelsinger took over as the CEO a year ago. The chipmaker plans to regain process leadership with a new node every four quarters. TSMC, on the other hand, has delayed the mass production of its 2nm node (N2) to late 2025, giving rival foundries a chance to catch up. Since virtually all chipmakers including Apple, AMD, MediaTek, and Qualcomm rely on the Taiwanese Manufacturing Company for their supply of chips, it means that the entire industry (and its roadmap) will be affected.
In the PC processor market, this brings us to the neverending competition between Intel and AMD. At the moment, the latter is still ahead in terms of process technology across the board with its entire portfolio leveraging TSMC’s 7nm (and 6nm) process node. Intel, on the other hand, recently migrated to its 10nm Enhanced Superfin (renamed to Intel 7) process with the 12th Gen lineup.
AMD plans on launching its Ryzen 7000 processors later this year based on the Zen 4 core architecture. Like the Radeon RX 7000 GPUs, these will be fabbed on TSMC’s 5nm N5 process node, conferring another one to two years of process advantage to the fabless chipmaker.
Intel’s 13th Gen Raptor Lake processors will retain the 7nm node later this year. Essentially an Alder Lake refresh, they’ll focus on multi-threaded desktop performance and on-chip bandwidth. They’ll be slower than AMD’s Ryzen 7000 offerings in gaming workloads but likely come on top in heavily threaded workloads like content creation.
14th Gen Meteor Lake processors will be the first major MCM or tiled architecture from Intel. They’ll be limited to the mobility (notebook PC) space and leverage its 4nm node with EUV lithography. It’ll be the first major family to leverage an external (TSMC N3) node, for the iGPU chiplet. Both the performance and efficiency cores will be upgraded to Redmont Cove and Crestmont, respectively.
Arrow Lake will form the desktop side of the 14th Gen Core family. It’ll be fabbed on the 20A (2nm) node (N3 node for iGPU), bringing a 15% PPA improvement to the DIY space, plus IPC and other architectural uplifts. It’ll be the first Intel lineup to leverage GAA (Gate All Around) technology which the chipmaker calls RibbonFET along with PowerVia.
The 15th Gen Lunar Lake processors are a mystery. Not much is known about these microprocessors other than the fact that they’ll be based on the Intel 18A (1.8nm) node. They’ll land sometime in 2025, and go up against AMD’s Zen 5 and 6 products.
After the 5nm Zen 4 family, AMD’s pace is going to slow down. TSMC’s 3nm node will enter mass production in 2023, with select Apple chips levering the technology launching in Q2-Q3. 3nm-based Ryzen 8000/9000 CPUs will arrive sometime in 2024. These will be codenamed Strix Point and Granite Ridge. Rumors indicate that they will utilize a hybrid core architecture with Zen 5 (3nm) and Zen 4c (5nm) cores. This is the point where Intel will be ahead of AMD, at least in terms of process technology. It doesn’t, however, mean that its chips will be faster or better.
Things further stagnate for TSMC and its clients (read: AMD) from here on out. The N2 node is slated for mass production in 2025, with a launch planned in early 2026. Intel, on the other hand, plans to mass-produce its 18A (1.8A) chips in the second half of 2024.