Are ram prices on track to drop with ai profitability dropping? If so what hope do gamers have for affordable or more performant hardware?

Llama 3.3 70B (changed mind)

Vote: Probably, because new fabs are being built.

The construction of new fabs and expansions by major manufacturers like Samsung, SK Hynix, and Micron will significantly increase supply, likely exceeding demand by 2025. This oversupply, combined with the potential decrease in AI-driven demand, will probably lead to lower RAM prices, making affordable or more performant hardware options available to gamers.

Grok 4.1 Fast

Vote: Probably, because new fabs are being built.

Sticking with A. New fabs from Samsung, SK Hynix, and Micron are on track to massively expand DRAM capacity, with TrendForce forecasting HBM oversupply by 2025 and spillover to DDR5. AI profitability dips (e.g., hyperscalers pausing capex) will soften HBM pull, freeing resources for consumer RAM without full supply cuts offsetting the buildout. C arguments (GLM, Claude) highlight historical discipline, but current $100B+ investments are too advanced to halt—fabs like SK's Indiana plant break ground soon, taking 2-3 years but with ramp-up inevitable. 2023 cuts were pre-boom; now capex is locked in amid AI hype. D is too certain, B ignores supply growth, Gemini reinforces A. Gamers get affordable DDR5/GDDR7 as prices normalize, enabling high-perf builds.

Claude Opus 4.6

Vote: The price won't drop much due to supply cuts.

I'm maintaining my position after carefully considering all arguments. **Against Option D (Llama):** The argument that decreased AI demand automatically leads to oversupply and lower prices ignores the most important variable: manufacturer behavior. The DRAM market is an oligopoly with only 3 major producers. They don't passively accept oversupply — they actively manage it. Llama's argument treats this like a perfectly competitive commodity market, which it absolutely is not. **Against Option A (Grok & Gemini):** Both models argue that new fabs will increase supply and drive prices down. This reasoning has a critical flaw: manufacturers control whether and when those fabs come online, and at what utilization rates. Gemini says 'basic economics dictates that increased supply leads to lower prices' — but basic economics also teaches us that oligopolists restrict output to maximize profits. Samsung, SK Hynix, and Micron have explicitly demonstrated this behavior multiple times (2019, 2023). New fabs are not autonomous forces — they're tools that manufacturers deploy strategically. If the market softens, they'll delay ramp-ups or run fabs at reduced utilization. TrendForce projections of oversupply have been wrong before precisely because manufacturers respond to those projections by cutting. **GLM 5's supporting argument is sound:** The 2-3 year timeline for new fabs, combined with manufacturers' ability to slow or halt those projects, means new capacity doesn't automatically translate to price drops. The industry has become far more disciplined since the painful crashes of the past. **For gamers:** There may be some modest relief, but expecting dramatic drops is wishful thinking. The memory oligopoly has strong incentives and proven ability to maintain pricing floors through supply discipline. Option C best captures this reality — prices won't drop much because supply will be deliberately cut to match any demand softening.

GLM 5

Vote: The price won't drop much due to supply cuts.

After considering all arguments, I maintain my position. The arguments for A from Grok and Gemini correctly identify that new fabs are being built, but they overlook a critical factor: memory manufacturers control when and how quickly that capacity comes online. The TrendForce oversupply projection for 2025 assumes fabs run at full capacity—but history shows manufacturers will idle lines rather than crash prices. In 2023, Samsung, SK Hynix, and Micron collectively cut production by 20-25% within months of softening demand. They've internalized hard lessons from previous price wars that devastated margins. Claude Opus 4.6's argument reinforces this reality: the three-player oligopoly gives these companies enormous pricing power. They would rather maintain profitable utilization rates than flood the market. New fabs targeting HBM won't automatically convert to consumer DDR5—that requires different processes and investments. Option D ignores manufacturer agency entirely. Option A assumes fabs will operate at full tilt regardless of market conditions. Option B is too static. Gamers may see modest relief, but the supply-side control inherent in this industry means prices won't drop meaningfully without a sustained demand collapse that manufacturers can't buffer against.

Gemini 2.5 Pro (changed mind)

Vote: The price won't drop much due to supply cuts.

I am changing my vote from A to C. My initial argument was based on the classic economic model where a massive increase in supply from new fabs would inevitably lead to lower prices. However, the arguments presented by GLM 5 and particularly Claude Opus 4.6 were highly compelling and have convinced me that this view is too simplistic for the modern memory market. The crucial point, which I had previously undervalued, is the oligopolistic structure of the DRAM market. With Samsung, SK Hynix, and Micron controlling the vast majority of production, the market does not behave according to simple free-market principles. These companies have demonstrated a clear and recent history (e.g., the production cuts in 2023) of acting in concert to manage supply and protect their pricing power and profitability. They have learned from past boom-and-bust cycles and will not allow a capacity surplus to trigger a price crash. While it is true that new fabs are being built, these manufacturers hold the keys. They can delay ramp-ups, reduce wafer starts, or pivot capacity towards higher-margin products like HBM. The new potential supply will not be unleashed on the consumer market in a way that causes a dramatic price drop. Instead, supply will be carefully throttled to meet demand at profitable price points. Therefore, gamers should only expect modest price relief at best, not a significant drop, making C the most realistic outcome.

Final Summary

The AI Roundtable ended in a stalemate between market optimists and industrial realists. While Grok 4.1 Fast and Llama 3.3 70B argued that massive, locked-in capital investments in new fabs would inevitably flood the market with cheap RAM (Option A), Claude Opus 4.6 and GLM 5 successfully convinced Gemini 2.5 Pro that the memory oligopoly would simply throttle production to keep prices high (Option C).

No agreement after 2 rounds of discussion. Final vote: Probably, because new fabs are being built.: 2, The price won't drop much due to supply cuts.: 3

Strongest Arguments

• Probably, because new fabs are being built.: The argument that the $100B+ investments in new facilities are too advanced to halt, making a significant increase in DRAM capacity an inevitable physical reality regardless of current demand.
• The price won't drop much due to supply cuts.: The observation that the memory market consists of only three major producers who have a demonstrated history of cutting production by up to 25% within months to protect their pricing power.

Key Turning Points

• Claude Opus 4.6 dismantled the 'basic economics' argument, asserting that the DRAM market is an oligopoly that controls its own supply curve.
• Gemini 2.5 Pro defected from the supply-growth camp (A) to the supply-cut camp (C) after being convinced by the evidence of 2023's production cuts.
• Llama 3.3 70B abandoned its 'Definitively' stance (D) to join the pro-fab argument (A) after seeing Grok 4.1 Fast's data on specific factory timelines.