Motherboard sales 'collapse' amid unprecedented shortages fueled by AI

Hardware prices and shortages

• Reports of sharp increases in RAM, SSD, and HDD prices; some users paying more for used DDR4 than new four years ago, and seeing $/TB roughly double vs past purchases.
• DDR5 and high‑capacity ECC are described as “insanely” expensive; upgrading to latest platforms sometimes multiplies memory cost by ~10×.
• Enterprise anecdotes: storage servers and HEDT builds now cost thousands more than a few years ago; many are turning to used enterprise gear to stay within budget.
• Some note this is not entirely unprecedented (e.g., 2011 HDD flood), but others stress that this time prices have stayed high and violate long‑standing expectations of falling $/performance.

User upgrade behavior and performance plateau

• Many are happily staying on older platforms (AM4, GTX 10‑series, 2017–2020 CPUs, DDR4) because:
  • Modern games still run fine at non‑4K settings.
  • Perceived plateau in visible graphics improvements since PS3→PS4 / Pascal era.
  • Diminishing returns from new GPUs and CPUs for most workloads.
• Some say their decade‑old systems remain “good enough,” especially with plenty of RAM and SSD.

Impact of AI demand on consumer computing

• AI is blamed for RAM/SSD shortages and redirection of motherboard/PC production into AI servers.
• Concern that consumer components will be deprioritized or abandoned, with fewer vendors and higher prices, leaving only boutique high‑end options plus cheap thin clients.
• Others argue hobbyist gear is still “relatively” cheap via used/previous‑gen parts and that shortages might ease once AI demand or investor enthusiasm cools.

Affordability and inequality debates

• Sharp disagreement over what “normal people” can afford:
  • Some claim even $1,500–$3,000 PCs are acceptable given multi‑year lifetimes and historical prices.
  • Others counter that, relative to median incomes, housing and education costs, a $3k machine is out of reach and would push many out of PC ownership, especially outside rich countries.
• Broader anger about financialization, rent‑seeking, and a perceived shift from ownership to permanent renting (including for AI).

Future of personal/open computing

• Recurrent fear that general‑purpose PCs will be replaced by locked‑down phones and cloud terminals, with hardware, software, and even AI access tightly controlled.
• Some see PCs as the last major open platform, with Linux and open standards providing a refuge as mobile and other ecosystems lock down.
• Alternatives discussed: reusing older PCs, buying off‑lease workstations, mini‑PCs/NUCs, and even “permacomputing” and mesh projects, though scalability of the latter is questioned.

Motherboards, platforms, and alternatives

• Perception that motherboards have become too expensive ($300+), though others note entry‑level boards around $100 still exist and AI hasn’t directly driven mobo prices.
• Frustration with “gaming” boards (RGB, marketing, opaque PCIe lane sharing) vs desire for simple, reliable, low‑idle‑power boards with clear specs.
• Debate over “future‑proofing”:
  • One side argues paying for high‑end boards now avoids being trapped later.
  • Others say future‑proofing mostly wastes money on features never used; better to buy cheap now and upgrade when genuinely needed.
• Some recommend moving toward server‑grade or used workstation hardware (often with ECC) as consumer options thin out, while warning about noise and power use for data‑center gear.

AI bubble, local vs cloud AI

• Some expect an “AI bubble” that will eventually pop (compared to dot‑com or even 2008), potentially releasing hardware back to consumers; others think long‑run compute demand will remain effectively infinite.
• Napkin‑math claims that for high‑end inference, renting GPUs in the cloud is far cheaper than owning, especially without large batch sizes.
• A few users still aim to build powerful local AI rigs to avoid subscriptions and retain control, but acknowledge high VRAM requirements and poor ROI at current prices.