Memory and storage stocks slid Wednesday after Google introduced TurboQuant, a new compression algorithm meant to lower memory needs for AI systems. The move came even as the broader technology-heavy Nasdaq 100 posted gains, highlighting a sector-specific market reaction.
The pullback saw SanDisk Corporation (NASDAQ:SNDK) fall 5%, Micron Technology (NASDAQ:MU) lose 4%, Western Digital (NASDAQ:WDC) decline 3.7% and Seagate Technology (NASDAQ:STX) slide 4% on the session.
Google described TurboQuant as a 3-bit compression technique for AI models. In response, Lynx Equity Strategies analyst KC Rajkumar pushed back on the narrative around the announcement in a client note, saying the concept was not as novel as some third-party technology blogs have portrayed.
"Vector quantization is hardly a new idea as LLM models went from 32-bit data in original training models to 4-bit data used in inference models today," Rajkumar wrote. He faulted coverage that suggested an 8x improvement, noting that such comparisons were being made to older 32-bit models rather than the 4-bit quantized inference models now commonly used.
"While there may be an improvement, it is not 8x, in our view," the analyst added.
Rajkumar acknowledged that AI infrastructure providers must continue to innovate to address scaling challenges as token context length expands during inferencing. Still, he argued that these advances in compression do not translate into a near-term collapse in demand for memory and flash, citing supply constraints as a key factor.
"Advanced compression techniques merely reduce bottlenecks without destroying demand for dram/flash," Rajkumar wrote, emphasizing that memory and flash requirements remain tied to broader capacity and supply dynamics over the coming three to five years.
Lynx Equity Strategies left its $700 price target on Micron intact and reiterated a buy recommendation for the stock. The analyst concluded, "We would be buyers into the Google-related pullback today."
The session's coverage also included reference to ProPicks AI, an evaluation tool that analyzes companies including Micron across numerous financial metrics. The original copy cited ProPicks AI's methodology and cited past winners identified by the tool, including Super Micro Computer (+185%) and AppLovin (+157%), as examples of its track record.
Where this matters
- Semiconductor and storage sectors are directly affected by developments in model compression and memory efficiency.
- AI infrastructure providers and cloud operators face incentives to optimize memory usage as model context lengths grow.
- Investor sentiment in technology sub-sectors can diverge from broader market performance, as seen with the Nasdaq 100 advancing while memory names fell.