Intel Nova Lake and Raptor Lake: The Confirmed Roadmap and the DDR5 Price Crisis Driving Two Generations in Parallel

Desktop-First Launch Strategy

Intel VP Robert Hallock confirmed that Nova Lake's new core architecture will launch on desktop first—a deliberate reversal of the typical server-first deployment. Hallock said he hopes enthusiasts "do the math" compared to AMD's approach, signaling Intel's bet that Nova Lake can reclaim desktop competitiveness against AMD's Zen 6 .

Staggered 2027 Rollout (Not 2026)

Intel CEO Lip-Bu Tan originally confirmed a late-2026 launch window . However, more recent reports from August 2026 indicate that Nova Lake-S will not ship in 2026. Leaked internal roadmaps show a four-wave rollout spanning most of 2027 :

• Wave 1 (late January – March 2027): 28-core single-compute-tile models launch first, with a CES 2027 announcement but no immediate retail availability .
• Wave 2 (Q2 2027): The 52-core flagship arrives later, featuring a big L3 cache ("bLLC") and a dual-compute-tile design .

Earlier reports from June 2026 by Tom's Hardware also noted that the first Nova Lake SKUs would roll out at CES 2027, with the 52-core flagship following around Computex .

Raptor Lake: A "Core Part of the Portfolio" for Years

Intel has explicitly confirmed that Raptor Lake (13th and 14th Gen) will remain in production for years, calling it a "core part of the portfolio" . The reason is a dramatic shift in memory pricing that has reversed the normal generational upgrade cycle.

The DDR5 Price Crisis

• A 32 GB DDR5-6000 kit cost under $90 in early 2025. By mid-2026, the cheapest kit cost over $375—a more than 4x increase .

Root Cause: AI Is Cannibalizing Consumer DRAM

Samsung, SK Hynix, and Micron—which together control roughly 70% of the DRAM market—have collectively redirected over 50% of DRAM wafer capacity to high-bandwidth memory (HBM) production for AI accelerators. HBM is far more profitable than consumer DDR5, structurally starving DDR4 and DDR5 commodity DRAM of manufacturing resources .

Demand Backflow and Intel's Response

Intel's Robert Hallock described a "sudden inrush of demand" for LGA 1700/DDR4 systems as builders fled expensive DDR5. Intel responded by committing to keep Raptor Lake "abundantly available" for years and validating lower memory configurations . Reports indicate Intel is even preparing Raptor Lake Next, a third refresh on LGA 1700 for H1 2027, still supporting DDR4 .

The result: Intel is running two parallel desktop strategies—Nova Lake (new socket, DDR5-only, high-end) alongside extended Raptor Lake production (LGA 1700, DDR4, value-oriented) .

Broader Industry Shift: Legacy Platforms Get a Second Life

Intel's situation is not an isolated case. The entire DRAM and semiconductor industry is experiencing a structural inversion driven by AI demand.

DDR4 Outlasting Its EOL

Samsung and SK Hynix extended DDR4 production through at least the end of 2026 . In 2025, DDR4 briefly became more profitable per-bit than DDR5—reversing the normal generational price curve . Samsung initially planned to wind down its 1z-node DDR4 production lines by late 2025 but postponed the shutdown due to strong demand and rising prices .

Memory Makers Prioritize AI Over PC

HBM and server DRAM command higher margins, so the big three memory manufacturers are accelerating DDR4 phase-outs even as demand for it remains strong, creating a shortage of "less glamorous" memory types . TrendForce reported that the three major DRAM suppliers are allocating advanced process capacity primarily to high-end server DRAM and HBM, crowding out capacity for PC, mobile, and consumer applications .

Platform Fragmentation

OEMs and chipmakers are forced to maintain DDR4 and DDR5 product lines simultaneously. Intel extends Raptor Lake; Samsung and SK Hynix keep DDR4 fabs running; motherboard makers produce both LGA 1700 and new-socket boards. AMD is also expected to maintain AM4/DDR4 support for budget segments .

The key driver: AI infrastructure demand is cannibalizing consumer DRAM supply. This has created an unusual market where legacy platforms are not just tolerated but actively necessary for cost-sensitive buyers, and chipmakers must maintain two generations of platforms far longer than originally planned .