IGN’s argument that AI is making PC gaming worse has one hard, defensible foundation: memory scarcity is now raising the cost of gaming hardware, while many of the consumer-facing AI features arriving alongside it remain unproven or unwanted. But the evidence points to a more precise problem than “AI” in general. Data-center demand is pulling memory manufacturing toward high-margin HBM and server DRAM, and gamers are seeing the consequences through tighter supply and higher prices for system RAM, GDDR graphics memory, and finished graphics cards.
Jackie Thomas’s August 1 column correctly connects the consumer squeeze to the AI infrastructure build-out. Reuters reported in December that Microsoft, Google, ByteDance, Amazon, Meta, and other large buyers were competing to lock up supplies from Micron, Samsung, and SK Hynix. It also reported that the shift toward high-end memory was constraining conventional DRAM for PCs and consumer devices. Samsung’s latest earnings reinforce that broad picture: its semiconductor business has been lifted by rising memory prices, AI-server demand, and high-bandwidth-memory shipments.
The important correction is that this is not a clean case of AI companies buying every ordinary RAM chip “before it exists,” nor is every higher-priced device a direct result of a chatbot or an image generator. It is a capacity-allocation problem. The same companies make memory products for AI accelerators, servers, PCs, phones, consoles, and cars; when limited fab capacity produces better returns in HBM and server memory, mainstream products lose priority. That distinction does not make the pressure on PC builders less real. It explains why it may endure longer than the cryptocurrency GPU shortages gamers remember.
There is now visible evidence that high-end graphics cards are getting more expensive in the United States. Tom’s Hardware found Asus ROG Astral RTX 5080 listings at $2,099.99 at Best Buy, $600 above that card’s launch MSRP, while certain RTX 5090 partner cards had climbed above $4,300. Those are retail prices for premium board-partner models, not a formal Nvidia MSRP increase—and the distinction matters.
Some less extravagant RTX 5080 models were still available closer to their original suggested prices. Tom’s Hardware also noted another complicating factor: reports that Nvidia’s observed-pricing incentive program for board partners had ended, removing one mechanism that encouraged cards to stay near MSRP. Memory costs are a credible contributor, but they are not the only plausible reason a halo Asus or MSI card suddenly costs hundreds more.
AMD’s situation is even less settled than the column suggests. TrendForce reported in July, citing Chinese supply-chain reporting, that AMD had notified add-in-board partners of an approximately 10% increase in GPU-and-GDDR bundle pricing. The report said that change took effect in July and attributed it to GDDR shortages, but AMD has not publicly announced such a change. It is reporting on an upstream component-bundle price, not a confirmed 10% hike for every Radeon card on a U.S. store shelf.
That leaves buyers with an unglamorous but useful conclusion: do not treat a reported supply-chain increase as a universal retail-price rule. Compare actual cards, not launch MSRPs from 2025 or a rumored percentage from a distributor channel. An RTX 5080 or Radeon RX 9070 XT may be a bad buy at one retailer and a reasonable one at another, even during a broad memory squeeze.
What is still missing is the disclosure gamers would need to calculate the impact cleanly. Samsung and SK Hynix have announced large investments, yet Reuters noted that they had not detailed the eventual production split between HBM and conventional memory. That means neither a PC builder nor a board partner can look at a public document and determine when DDR5, GDDR6, or GDDR7 supply will normalize.
The timeframe is also longer than a typical retail restock cycle. New fabs take years, not weeks, and SK Hynix’s leadership has warned that the shortfall could extend well into 2027. Samsung has likewise said supply constraints may remain severe. More capacity is coming, but it will arrive into a market where AI infrastructure buyers are still expanding. Price relief for desktop memory and graphics cards is therefore not guaranteed merely because fabs are being announced.
For Windows gamers, this changes the value calculation around upgrades. A CPU platform move that requires new DDR5 kits, a motherboard, and more storage may now have a far worse cost-to-performance ratio than a targeted GPU upgrade—or no upgrade at all. Existing systems with 16GB of RAM deserve more attention, too: memory pressure is making the longstanding advice to move to 32GB more expensive precisely as modern games, browsers, launchers, and Windows background workloads make that capacity more attractive.
Nvidia’s March announcement for DLSS 5 moves past that relatively narrow role. Nvidia describes DLSS 5 as a real-time neural-rendering model that takes a game’s color and motion vectors and uses AI to add lighting and material detail while anchoring the result to the source 3D scene. Nvidia says artists will have controls for intensity, color grading, and masking, and it has promised support in games including Resident Evil Requiem, Starfield, Assassin’s Creed Shadows, and The Elder Scrolls IV: Oblivion Remastered.
That is why the negative reaction to Nvidia’s Resident Evil Requiem demonstration cannot be waved away as resistance to upscaling. DLSS 5 is intended to alter the final visual presentation, not simply reconstruct pixels from a lower internal resolution. Nvidia calls that a route to photorealistic materials and lighting; critics see a system that can overwrite deliberate character art, texture work, and scene composition with a model’s statistical guess about what looks “real.”
Nvidia’s announcement is still a promise, not a shipping verdict. DLSS 5 is scheduled to arrive this fall, so there is no public retail build that establishes its quality, GPU requirements, latency costs, per-game controls, or whether players can fully disable it without losing other rendering features. The company’s own launch material recognizes the issue indirectly by emphasizing artistic controls. Until games ship, the real question is not whether neural rendering can produce impressive screenshots. It is whether developers retain enough control to keep it from homogenizing a game’s visual identity.
That feature is optional, and the hardware shortage is not the price of a Copilot query. Still, the product decision is revealing. Windows handhelds already demand better controller navigation, launcher integration, driver stability, low-power tuning, shader-cache management, and clear performance diagnostics. The ROG Xbox Ally X reviews from Ars Technica and WIRED found genuine progress in the Xbox full-screen experience and Game Bar, while also noting Microsoft’s determination to insert Copilot into the PC-gaming experience.
Microsoft’s own announcement described Gaming Copilot as a beta. It did not publish accuracy measurements for game-specific advice, a supported-game list, or a guarantee that answers will be current after a patch changes a quest, item, build, or boss encounter. A player can disable or ignore a poor companion overlay. A player cannot so easily avoid the price of GDDR memory embedded in a graphics card.
That is a workable dividing line. It does not declare that every developer who uses a code helper has made an “AI game.” It gives a buyer information when generated art, writing, audio, or live content reaches the product or its marketing.
Epic CEO Tim Sweeney has dismissed game-store AI labels as pointless because AI will become common throughout development. Yet ubiquity is not a reason to remove disclosure. It is the reason to make disclosure more specific. A game using AI for internal localization drafts has a different consumer implication from a game selling AI-generated character portraits, voice performances, item art, or dialogue. Steam’s policy does not resolve authorship, quality, copyright, or labor concerns, but it gives buyers a chance to make that distinction before purchase.
The sharper conclusion from IGN’s column is not that every AI technique has made games worse. AI upscaling remains valuable where it delivers performance without overriding a developer’s work; Windows gaming assistance may be useful when it is accurate and stays out of the way. The industry’s failure is treating a constrained memory supply and a player’s trust in the image on screen as expendable inputs while prioritizing the highest-margin AI opportunity.
For PC gamers, the immediate consequence is practical: delay nonessential RAM purchases if pricing is irrational, price-check actual GPU models instead of assuming MSRP is meaningful, and scrutinize whether an AI-branded graphics feature is solving a performance problem or simply changing the game you bought.
The important correction is that this is not a clean case of AI companies buying every ordinary RAM chip “before it exists,” nor is every higher-priced device a direct result of a chatbot or an image generator. It is a capacity-allocation problem. The same companies make memory products for AI accelerators, servers, PCs, phones, consoles, and cars; when limited fab capacity produces better returns in HBM and server memory, mainstream products lose priority. That distinction does not make the pressure on PC builders less real. It explains why it may endure longer than the cryptocurrency GPU shortages gamers remember.
The GPU price story is real, but the vendor-price claim is still thin
There is now visible evidence that high-end graphics cards are getting more expensive in the United States. Tom’s Hardware found Asus ROG Astral RTX 5080 listings at $2,099.99 at Best Buy, $600 above that card’s launch MSRP, while certain RTX 5090 partner cards had climbed above $4,300. Those are retail prices for premium board-partner models, not a formal Nvidia MSRP increase—and the distinction matters.Some less extravagant RTX 5080 models were still available closer to their original suggested prices. Tom’s Hardware also noted another complicating factor: reports that Nvidia’s observed-pricing incentive program for board partners had ended, removing one mechanism that encouraged cards to stay near MSRP. Memory costs are a credible contributor, but they are not the only plausible reason a halo Asus or MSI card suddenly costs hundreds more.
AMD’s situation is even less settled than the column suggests. TrendForce reported in July, citing Chinese supply-chain reporting, that AMD had notified add-in-board partners of an approximately 10% increase in GPU-and-GDDR bundle pricing. The report said that change took effect in July and attributed it to GDDR shortages, but AMD has not publicly announced such a change. It is reporting on an upstream component-bundle price, not a confirmed 10% hike for every Radeon card on a U.S. store shelf.
That leaves buyers with an unglamorous but useful conclusion: do not treat a reported supply-chain increase as a universal retail-price rule. Compare actual cards, not launch MSRPs from 2025 or a rumored percentage from a distributor channel. An RTX 5080 or Radeon RX 9070 XT may be a bad buy at one retailer and a reasonable one at another, even during a broad memory squeeze.
HBM demand is affecting gaming, but no manufacturer has explained the full split
Reuters’ reporting provides the strongest account of the underlying shortage. DRAM supplier inventories fell sharply during 2025, technology companies sought open-ended supply commitments, and Samsung and SK Hynix expanded capacity while cautioning that new conventional-memory output would not appear quickly. Samsung and SK Hynix have since posted record financial results tied to the AI-memory boom.What is still missing is the disclosure gamers would need to calculate the impact cleanly. Samsung and SK Hynix have announced large investments, yet Reuters noted that they had not detailed the eventual production split between HBM and conventional memory. That means neither a PC builder nor a board partner can look at a public document and determine when DDR5, GDDR6, or GDDR7 supply will normalize.
The timeframe is also longer than a typical retail restock cycle. New fabs take years, not weeks, and SK Hynix’s leadership has warned that the shortfall could extend well into 2027. Samsung has likewise said supply constraints may remain severe. More capacity is coming, but it will arrive into a market where AI infrastructure buyers are still expanding. Price relief for desktop memory and graphics cards is therefore not guaranteed merely because fabs are being announced.
For Windows gamers, this changes the value calculation around upgrades. A CPU platform move that requires new DDR5 kits, a motherboard, and more storage may now have a far worse cost-to-performance ratio than a targeted GPU upgrade—or no upgrade at all. Existing systems with 16GB of RAM deserve more attention, too: memory pressure is making the longstanding advice to move to 32GB more expensive precisely as modern games, browsers, launchers, and Windows background workloads make that capacity more attractive.
DLSS 5 is a different argument from DLSS upscaling
IGN is also right to separate AI-assisted performance technology from generative visual processing. DLSS began as an AI upscaling technology and later added frame generation; those tools can reduce the amount of conventional rendering a GPU must perform. Players can debate image quality, input latency, and marketing claims, but the technical purpose is direct: make demanding games playable at better settings or frame rates.Nvidia’s March announcement for DLSS 5 moves past that relatively narrow role. Nvidia describes DLSS 5 as a real-time neural-rendering model that takes a game’s color and motion vectors and uses AI to add lighting and material detail while anchoring the result to the source 3D scene. Nvidia says artists will have controls for intensity, color grading, and masking, and it has promised support in games including Resident Evil Requiem, Starfield, Assassin’s Creed Shadows, and The Elder Scrolls IV: Oblivion Remastered.
That is why the negative reaction to Nvidia’s Resident Evil Requiem demonstration cannot be waved away as resistance to upscaling. DLSS 5 is intended to alter the final visual presentation, not simply reconstruct pixels from a lower internal resolution. Nvidia calls that a route to photorealistic materials and lighting; critics see a system that can overwrite deliberate character art, texture work, and scene composition with a model’s statistical guess about what looks “real.”
Nvidia’s announcement is still a promise, not a shipping verdict. DLSS 5 is scheduled to arrive this fall, so there is no public retail build that establishes its quality, GPU requirements, latency costs, per-game controls, or whether players can fully disable it without losing other rendering features. The company’s own launch material recognizes the issue indirectly by emphasizing artistic controls. Until games ship, the real question is not whether neural rendering can produce impressive screenshots. It is whether developers retain enough control to keep it from homogenizing a game’s visual identity.
Gaming Copilot shows the cost of putting AI where players did not ask for it
Microsoft’s Gaming Copilot illustrates the other half of IGN’s complaint: AI can be intrusive even when it has no material connection to semiconductor pricing. Microsoft began expanding the beta through Game Bar on Windows PCs in September 2025 and positioned it as help for players stuck on bosses, looking for recommendations, or seeking in-game guidance. It supports push-to-talk and screen-aware assistance.That feature is optional, and the hardware shortage is not the price of a Copilot query. Still, the product decision is revealing. Windows handhelds already demand better controller navigation, launcher integration, driver stability, low-power tuning, shader-cache management, and clear performance diagnostics. The ROG Xbox Ally X reviews from Ars Technica and WIRED found genuine progress in the Xbox full-screen experience and Game Bar, while also noting Microsoft’s determination to insert Copilot into the PC-gaming experience.
Microsoft’s own announcement described Gaming Copilot as a beta. It did not publish accuracy measurements for game-specific advice, a supported-game list, or a guarantee that answers will be current after a patch changes a quest, item, build, or boss encounter. A player can disable or ignore a poor companion overlay. A player cannot so easily avoid the price of GDDR memory embedded in a graphics card.
Steam’s disclosure rule is more useful than Tim Sweeney allows
The column’s concern over AI-generated game art also deserves more than a cultural argument. Valve tightened Steam’s generative-AI disclosure form in January. As Game Developer reported, developers do not have to disclose ordinary AI-powered workflow tools such as coding assistance, but they must disclose generative AI used to create game content or associated material, as well as AI-generated content produced during gameplay.That is a workable dividing line. It does not declare that every developer who uses a code helper has made an “AI game.” It gives a buyer information when generated art, writing, audio, or live content reaches the product or its marketing.
Epic CEO Tim Sweeney has dismissed game-store AI labels as pointless because AI will become common throughout development. Yet ubiquity is not a reason to remove disclosure. It is the reason to make disclosure more specific. A game using AI for internal localization drafts has a different consumer implication from a game selling AI-generated character portraits, voice performances, item art, or dialogue. Steam’s policy does not resolve authorship, quality, copyright, or labor concerns, but it gives buyers a chance to make that distinction before purchase.
The sharper conclusion from IGN’s column is not that every AI technique has made games worse. AI upscaling remains valuable where it delivers performance without overriding a developer’s work; Windows gaming assistance may be useful when it is accurate and stays out of the way. The industry’s failure is treating a constrained memory supply and a player’s trust in the image on screen as expendable inputs while prioritizing the highest-margin AI opportunity.
For PC gamers, the immediate consequence is practical: delay nonessential RAM purchases if pricing is irrational, price-check actual GPU models instead of assuming MSRP is meaningful, and scrutinize whether an AI-branded graphics feature is solving a performance problem or simply changing the game you bought.
References
- Primary source: in.ign.com
Published: 2026-08-01T16:00:00+00:00
- Related coverage: nvidianews.nvidia.com
NVIDIA DLSS 5 Delivers AI-Powered Breakthrough in Visual Fidelity for Games | NVIDIA Newsroom
NVIDIA today unveiled NVIDIA DLSS 5, the company’s most significant breakthrough in computer graphics since the debut of real-time ray tracing in 2018.nvidianews.nvidia.com