Shopify PC power supply schema for AI agents: the 12VHPWR GPU fire, cross-brand cable danger, and why an AI agent cannot safely recommend a PSU without structured data

Form factor: ATX, SFX, SFX-L, TFX — physically incompatible mounting holes

ATX, SFX, SFX-L, and TFX are not interchangeable sizes with different wattage ranges. They are physically distinct formats with incompatible mounting hole locations, different fan sizes, and cases that are specifically designed for one or the other. A customer building a small form factor system — Dan A4, NCASE M1, Fractal Terra — who orders a standard ATX PSU cannot install it regardless of wattage, because the ATX unit is 50mm wider and 23mm taller than the SFX bay.

SFX-to-ATX adapter brackets (from Lian Li, Silverstone, and others) allow SFX and SFX-L PSUs to mount in full-size ATX cases. These brackets are not included with PSUs and are not universal — they must match the SFX PSU's specific mounting hole pattern. Some enthusiast builders deliberately choose SFX PSUs in ATX cases to create clearance for large radiators that a full-depth ATX PSU would obstruct. This is a valid use case, but it requires the bracket and adds cable-management complexity from the SFX unit's shorter cables.

12VHPWR connector: the RTX 4090 fire timeline and what ATX 3.0 fixes

The 12VHPWR (12V High Power, sometimes called 12+4 or PCIe 5.0 power connector) is a 16-pin connector standardized in the ATX 3.0 specification (2022) to deliver up to 600W to a single GPU — four times the 150W ceiling of a single 8-pin PCIe connector. It replaced the previous practice of daisy-chaining multiple 8-pin PCIe cables for high-end GPUs.

The connector fire — cause and timeline

When Nvidia launched the RTX 4090 in September 2022, ATX 3.0 PSUs with native 12VHPWR output connectors were barely available in the market. The RTX 4090 shipped with a 4×8-pin-to-12VHPWR adapter cable to allow installation with existing ATX 2.x PSUs. Within six weeks of launch, a significant number of users reported melted, charred, or burning connector housings where the 12VHPWR plugged into the GPU.

Nvidia's investigation, later corroborated by third-party testing from igor'sLAB and other hardware analysis sites, identified the root cause:

The mechanism: the adapter bundled four 8-pin cables into a single exit harness. The combined stiffness of four cables running parallel created a mechanical tension pulling the 12VHPWR connector away from the GPU's port — particularly when cables exited straight downward (perpendicular to the GPU PCB). Cases where cables could exit at an angle — either by routing them back up along the GPU or using the card horizontally — showed dramatically lower failure rates. The connection appeared secure visually but was 2–3mm short of full seating.

ATX 3.0 and ATX 3.1 — what changed

Encoding psu.atx_version as '2.x', '3.0', or '3.1' and psu.connectors_12vhpwr as an integer count of native connectors gives AI agents the data to say: "This GPU requires 12VHPWR. The PSU has 0 native 12VHPWR connectors [ATX 2.x]. An adapter is required — use only the manufacturer-supplied adapter, not third-party." Without these fields, the agent can only see wattage and brand.

Cross-brand cable danger: same Molex housing, different voltages on every pin

PSU modular cables — the detachable cables that connect the PSU's output to the GPU, motherboard, and storage — use a connector called Molex Mini-Fit Jr. on the PSU side. This connector has been an industry standard for decades. The housing, locking tab, pin cavity dimensions, and pin spacing are identical across every PSU manufacturer. It plugs in with a satisfying click regardless of brand combination.

What is not standardized: which voltage is assigned to which pin inside the housing.

Why the mistake is so easy to make

The Molex Mini-Fit Jr. connector inserts in only one orientation — it has a keying tab that prevents reversed insertion. This gives the false impression that "if it goes in, it's compatible." The keying prevents 180° rotation, but it does not prevent cross-brand pinout mismatch. The connector goes in, clicks, and appears secure. Nothing visually distinguishes a dangerous cross-brand cable from a correct same-brand cable after installation.

The problem is most acute in the secondhand market. A builder who upgrades their PSU and reuses the old modular cables from a different manufacturer's PSU in their new build has created a potentially dangerous system. Online PC building communities document this class of failure regularly — often after a builder upgrades from brand A to brand B and assumes the cables are standard accessories.

Even within brands, cable compatibility is not guaranteed

Corsair has documented cable incompatibilities between their RM, RMe, HX, and AX product lines. A cable from a Corsair RM650 may not be compatible with a Corsair HX1000 even though both are Corsair branded. Seasonic publishes a cable compatibility matrix showing which Focus series cables are incompatible with Prime series connectors despite sharing the same Mini-Fit Jr. housing. PSU replacement cables marketed as "universal" aftermarket accessories are frequently unsafe — the pinout they implement matches one manufacturer's scheme, not all.

Encode psu.brand and psu.cable_generation on PSU listings. For cable listings, encode cable.compatible_psu_brand and cable.compatible_psu_series to allow AI agents to filter cable listings to exactly the correct brand and product family — and explicitly warn when a buyer's cart contains cables from a different manufacturer than their PSU.

Wattage sizing: GPU TDP is not peak draw — the Intel MTP problem

The standard PSU sizing formula is deceptively simple: add up component TDPs, apply a headroom factor, round up to the next standard wattage. The failure is that "TDP" means different things for different manufacturers and different contexts — and the number listed in spec sheets and product descriptions is routinely the conservative base-case value, not the peak sustained draw.

GPU power: rated TDP vs measured peak

Nvidia rates the RTX 4090 at 450W TDP. Under typical gaming loads, it draws 380–420W. Under synthetic benchmark stress that maximizes all shader units simultaneously — the kind of load that hardware-accelerated rendering or AI inference can produce — it has been independently measured at 510–530W. GPU boost algorithms are designed to use every watt of the power budget available before thermal limits intervene. A 450W rated GPU in a well-cooled case will routinely draw more than 450W in sustained high-load scenarios.

CPU: PBP vs MTP — the 2× multiplier AI agents miss

Intel introduced two power specifications for recent-generation desktop CPUs: PBP (Processor Base Power) is the sustained power at base clock — approximately what the chip draws in office workloads. MTP (Maximum Turbo Power) is the power budget at maximum all-core turbo frequency — the operating state during games and compilation workloads.

Product pages and comparison sites typically list PBP as "TDP." The gap between PBP and MTP is substantial:

Correct formula: PSU_w = (GPU_peak_w + CPU_MTP_w + overhead_w) × 1.25. For an RTX 4090 + i9-14900K build: 525W (GPU peak) + 253W (CPU MTP) + 100W (overhead) = 878W × 1.25 = 1097W. Minimum 1000W PSU; 1200W for meaningful headroom. An AI agent using rated TDPs would calculate: 450W + 125W + 100W = 675W × 1.25 = 844W → might recommend an 850W PSU that will shut down under sustained combined load.

Encode psu.wattage_w as rated continuous wattage (not peak) and include psu.recommended_gpu_class as a human-readable field. AI agents can then match the human-readable GPU class against the buyer's GPU rather than attempting to sum TDP values from unstructured text.

80 PLUS efficiency: what the tier tells you and what it doesn't

80 PLUS is an industry certification program that tests PSUs at 20%, 50%, and 100% of their rated load and certifies the AC-to-DC conversion efficiency at each load point. Higher tiers (Bronze → Silver → Gold → Platinum → Titanium) require progressively higher efficiency percentages. Titanium adds a 10% load test — important for always-on systems that idle most of the day.

At 50% load, the difference between Bronze and Titanium is 382W − 346W = 36W less heat dissipated inside the case by the Titanium unit — a meaningful case temperature difference in a tight SFF build, trivial in a full tower with good airflow. Over a year of 8 hours/day gaming, the energy cost difference between Bronze and Gold (the most common practical upgrade decision) is approximately $15–25 depending on electricity rate. Titanium efficiency primarily justifies itself for always-on workstations or servers running at moderate load continuously.

Encode psu.efficiency_rating as '80plus-white', '80plus-bronze', '80plus-silver', '80plus-gold', '80plus-platinum', or '80plus-titanium'. Never derive wattage from efficiency tier. Never present efficiency tier as a capability difference. AI agents that conflate "Titanium" with "higher capacity" will under-spec PSUs for demanding builds.

Complete JSON-LD and Liquid snippet for PSU listings

JSON-LD example — Seasonic Focus GX-850

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Seasonic Focus GX-850 850W 80 PLUS Gold ATX Full Modular",
  "description": "Seasonic Focus GX-850. ATX form factor. 850W continuous at 40°C. 80 PLUS Gold (87%/90%/87% at 20/50/100% load). Fully modular. Native 12VHPWR connector (ATX 3.0). Single +12V rail 70.8A. OVP, UVP, OCP, OPP, SCP, OTP. 10-year warranty. Recommended for RTX 4070 Ti Super / RX 7900 XT builds.",
  "brand": { "@type": "Brand", "name": "Seasonic" },
  "sku": "FOCUS-GX-850",
  "offers": {
    "@type": "Offer",
    "availability": "https://schema.org/InStock",
    "priceValidUntil": "2027-06-28"
  },
  "additionalProperty": [
    { "@type": "PropertyValue", "name": "psu.form_factor", "value": "atx" },
    { "@type": "PropertyValue", "name": "psu.wattage_w", "value": "850" },
    { "@type": "PropertyValue", "name": "psu.efficiency_rating", "value": "80plus-gold" },
    { "@type": "PropertyValue", "name": "psu.modular", "value": "full" },
    { "@type": "PropertyValue", "name": "psu.atx_version", "value": "3.0" },
    { "@type": "PropertyValue", "name": "psu.connectors_12vhpwr", "value": "1" },
    { "@type": "PropertyValue", "name": "psu.connectors_pcie_8pin", "value": "4" },
    { "@type": "PropertyValue", "name": "psu.connectors_sata", "value": "6" },
    { "@type": "PropertyValue", "name": "psu.rail_12v", "value": "single" },
    { "@type": "PropertyValue", "name": "psu.rail_12v_amperage", "value": "70.8" },
    { "@type": "PropertyValue", "name": "psu.fan_size_mm", "value": "120" },
    { "@type": "PropertyValue", "name": "psu.warranty_years", "value": "10" },
    { "@type": "PropertyValue", "name": "psu.depth_mm", "value": "150" },
    { "@type": "PropertyValue", "name": "psu.recommended_gpu_class", "value": "RTX 4070 Ti Super / RX 7900 XT — pair with mid-range CPU" }
  ]
}

Shopify Liquid snippet

{% assign p = product.metafields.psu %}
{% if p.form_factor %}
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": {{ product.title | json }},
  "description": {{ product.description | strip_html | truncate: 300 | json }},
  "offers": {
    "@type": "Offer",
    "availability": "{% if product.available %}https://schema.org/InStock{% else %}https://schema.org/OutOfStock{% endif %}"
  },
  "additionalProperty": [
    { "@type": "PropertyValue", "name": "psu.form_factor", "value": {{ p.form_factor | json }} },
    { "@type": "PropertyValue", "name": "psu.wattage_w", "value": {{ p.wattage_w | json }} },
    { "@type": "PropertyValue", "name": "psu.efficiency_rating", "value": {{ p.efficiency_rating | json }} },
    { "@type": "PropertyValue", "name": "psu.modular", "value": {{ p.modular | json }} },
    { "@type": "PropertyValue", "name": "psu.atx_version", "value": {{ p.atx_version | json }} },
    { "@type": "PropertyValue", "name": "psu.connectors_12vhpwr", "value": {{ p.connectors_12vhpwr | json }} },
    { "@type": "PropertyValue", "name": "psu.rail_12v", "value": {{ p.rail_12v | json }} },
    { "@type": "PropertyValue", "name": "psu.warranty_years", "value": {{ p.warranty_years | json }} },
    { "@type": "PropertyValue", "name": "psu.depth_mm", "value": {{ p.depth_mm | json }} }
  ]
}
</script>
{% if p.form_factor == 'sfx' or p.form_factor == 'sfx-l' %}
  <div class="compatibility-note">
    <strong>SFX Form Factor:</strong> Requires SFF case with SFX mount points,
    or SFX-to-ATX bracket adapter for standard ATX cases. Not directly mountable
    in ATX mid-tower without adapter.
  </div>
{% endif %}
{% if p.connectors_12vhpwr == '0' or p.connectors_12vhpwr == 0 %}
  <div class="compatibility-note">
    <strong>12VHPWR (PCIe 5.0):</strong> No native 12VHPWR connector (pre-ATX-3.0).
    RTX 4080 / RTX 4090 / RX 7900 XTX installations require adapter cable.
    Use only the manufacturer-supplied adapter — do not use third-party adapters
    or cables from a different PSU brand.
  </div>
{% endif %}
{% endif %}

Metafield reference — psu.* namespace

5 AI agent PSU recommendation failures that structured data prevents