DRAM pricing is killing the hobbyist SBC market

Original Article ↗ Hacker News Discussion ↗

Soaring DRAM and flash prices driven by AI datacenter demand are pushing hobbyist single-board computers like the Raspberry Pi out of the low-cost range that made them so popular, with high‑RAM models now rivaling refurbished laptops and mini‑PCs in price. Commenters describe shifting to used x86 hardware and microcontrollers, worry about knock‑on effects on budget smartphones and other consumer gear, and debate whether this is a temporary bubble or a longer‑term structural change in how memory manufacturers allocate scarce capacity. Many also see a silver lining or necessity in renewed focus on efficient software and reusing older hardware rather than assuming ever‑cheaper RAM.

Impact on SBC Pricing and Use

  • Commenters report steep price hikes on higher-RAM SBCs (e.g., 8–16GB Raspberry Pi 5 and Compute Modules), in some cases nearing or exceeding cheap used laptops/mini PCs.
  • Many say that at current prices SBCs lose their “cheap tiny computer” appeal and are no longer obvious starter hardware for hobbyists.
  • Some argue “killing the market” is overstated: low‑RAM models (Zero 2 W, Pi 3 series, 2–4GB boards) remain relatively affordable, and past spikes (e.g., COVID) eventually eased.
  • Others stress that for low‑income and non‑US users, these price shifts are a real barrier to entry.

Shift to Microcontrollers and Used x86 Hardware

  • Rising SBC costs push people toward:
    • Microcontrollers (RP2040/Pi Pico, ESP32, Arduino, etc.), especially now that MicroPython/JS and UF2 flashing make them easier.
    • Used corporate SFF/mini PCs and laptops, which often offer 8–32GB RAM and SSDs for similar or less money, at the cost of higher power draw and no native GPIO.
  • Debate: some see SBCs as overkill for LED‑blinking‑type tasks; others value full Linux, TCP/IP, and familiar tooling as a learning bridge into embedded.

DRAM, Storage, and AI Demand

  • Multiple anecdotes of DRAM, SSD, HDD, and SD card prices jumping 3–10x, with server/workstation quotes 2–3x or more versus a few years ago.
  • Many attribute this primarily to AI datacenter demand and a shift of DRAM makers toward high‑margin HBM; others also mention fuel, helium supply, and geopolitics.
  • There is controversy over alleged “collusion” and reports of very large DRAM allocations to major AI firms; some see ordinary market response to big customers, others note past DRAM price‑fixing and risks of over‑reliance on one buyer.
  • Some expect new fabs and Chinese entrants to ease prices in a few years; others fear a longer‑term “new normal.”

Broader Consequences and Software Efficiency

  • Analysts cited predict that elevated memory costs will hit low‑ and mid‑range smartphones hardest, potentially halving volumes; hobbyist SBCs and budget phones are portrayed as “collateral damage” of the AI boom.
  • Many commenters lament software bloat and hope high RAM prices will force better memory discipline; others say orgs don’t care enough to prioritize this.
  • LLMs are seen both as part of the problem (driving demand) and as a possible tool to help write more efficient code, especially in languages with strong safety/type systems; there is significant skepticism that “just ask AI to optimize it” will work reliably.