RAM Companies' Stock Price Plummet, RAM Prices Go Down, But...

TL;DR

RAM prices dipped slightly and chip stocks crashed after Google's TurboQuant algorithm and rumors that OpenAI's 40% DRAM deal was non-binding letters of intent.

Key Points

• 1.RAM prices have dropped noticeably but remain far above historical lows. A Corsair Vengeance 32GB DDR5 kit fell from $543 to $350 at Best Buy, but that same kit once sold for as low as $87 — illustrating how far prices still are from pre-AI levels.
• 2.Google's TurboQuant algorithm triggered the stock sell-off. Announced March 24, 2026, the compression algorithm reduces LLM key-value cache memory by 6x with zero accuracy loss, causing investors to pull out of Micron, SK Hynix, and Samsung on fears of reduced AI RAM demand.
• 3.Analysts warn TurboQuant is evolutionary, not revolutionary. Ben Behringer of Quartier Chevier told CNBC the innovation 'does not alter the industry's long-term demand picture,' arguing that more efficient AI simply enables more powerful models that will require even more RAM.
• 4.OpenAI's deal for 40% of global DRAM supply may have been non-binding letters of intent. Analyst Akash Gupta claims Sam Altman signed simultaneous non-binding deals with Samsung and SK Hynix for 900,000 wafers/month in October 2025 — neither company knowing about the other — causing a 171% spike in contract DRAM prices and DDR5 kits jumping from $190 to $700.
• 5.Micron's consumer exit is now backfiring badly. Micron killed its Crucial consumer brand in December 2025 to redirect all wafers to AI customers, but is now down 33% from its post-earnings high despite revenue up 196% YoY and EPS up 682%, having restructured around demand signals from non-binding letters.
• 6.OpenAI is retreating from building data centers in favor of renting cloud capacity. Oracle and OpenAI scrapped the Stargate expansion site, OpenAI is renting Amazon Web Services instead of Microsoft Azure (risking a Microsoft lawsuit), signaling less direct RAM purchasing — though Oracle's broader 4.5 gigawatt data center commitment for OpenAI continues.
• 7.The broader AI spending race is financially damaging with no guaranteed payoff. Oracle fired 30,000 employees (18% of workforce) despite record revenues to fund AI infrastructure, massive AI investment contributed near-zero to US economic growth last year, and analysts broadly question whether OpenAI's business model is sustainable before a projected 2030 profitability target.