Iren (IREN) Canada GPU sites called number one worst in industry by users, citing multi-day power outages, storage failures, network issues and GPU errors, reports Semi Analysis

SemiAnalysis reports of this kind have become a recurring catalyst in the AI infrastructure trade: its site-level audits of neocloud operators have, in past episodes, produced sharp single-name moves precisely because the market has limited independent visibility into GPU fleet quality, uptime and networking performance, making third-party teardowns a price-setting input rather than background noise. The charges here go to the core of the business model, since GPU cloud revenue is underwritten by service-level uptime commitments, and multi-day outages, storage failures and network issues are exactly the failure modes that trigger SLA credits, contract non-renewals and, in severe cases, customer migration to competing capacity. The actors' track record matters: SemiAnalysis has established form as a technically credible, industry-followed source whose prior infrastructure assessments have been taken seriously by both equity holders and hyperscale counterparties, which raises the bar for a dismissal response from the company. What distinguishes this from a routine negative note is that the claims are operationally falsifiable, either outages occurred at the scale alleged or they did not, so the follow-ons to watch are the company's formal rebuttal, any customer commentary from named tenants, and whether the criticism is isolated to the named Canadian sites or extends to the broader fleet. For a name whose equity case rests on capacity credibility with large AI buyers, precedent in this cohort is that unresolved uptime questions weigh on contract wins and on the multiple applied to infrastructure revenue until answered with data rather than statements.