IREN said on 30 October that, “Under a multi-year agreement, Perplexity will access NVIDIA Blackwell Ultra GPUs at an IREN data center campus”. In a post on X the same evening, IREN said the workloads will support “the models and agents behind Perplexity Computer”.
IREN’s 27 August results release put its contracted annualized run-rate revenue for 2026 capacity at $4 billion, with $1 billion operating as of 26 August. By our arithmetic, $3 billion, or 75% of the contracted figure, was signed but not yet running on that date.
Business Pill · NEOCLOUDS EXPLAINED
A short Business Engineer explainer on neoclouds. It teaches the general idea only and does not discuss IREN or Perplexity.
The key insight: IREN says Perplexity will run Perplexity Computer workloads on Blackwell Ultra GPUs at an IREN campus under a multi-year agreement, and IREN’s own August figures put $4 billion of run rate under contract against $1 billion operating, a $3 billion gap by our arithmetic.
IREN’s blog post says Perplexity is “partnering with IREN to run AI workloads on IREN AI Cloud” and that “More than 50,000 organizations use Perplexity for accurate research-grounded AI”. IREN’s X post adds that Perplexity users ask “more than 1.5 billion questions a month”.
Nate Kupp, VP of Engineering at Perplexity, is quoted in the post: “Dedicated compute is one element of making that possible, and we’re excited to work with IREN as we invest in the next phase of Perplexity’s growth”. Tim Delcourt, Commercial Director, AI Cloud at IREN, says “At IREN, our job is to keep capacity ahead of AI ambition”.
The blog’s link on the word agreement points to IREN’s 20 July release, which listed Perplexity among its customers when IREN announced $2.8 billion of new multi-year contracts. IREN’s 27 August release named Perplexity among “Other recent signings”.
IREN stated $4 billion of contracted annualized run-rate revenue for 2026 capacity and $1 billion operating as of 26 August 2026 (FY26 results release, 27 August 2026). The $3 billion gap and the 75% share are our arithmetic. ARR is IREN’s operating metric, not GAAP revenue.
What Was Posted on X
Perplexity is working with IREN to power fast, trustworthy AI.
Under our multi-year agreement, @perplexity_ai will run workloads on NVIDIA Blackwell Ultra GPUs at an IREN data center campus, supporting the models and agents behind Perplexity Computer.
More than 50,000 organizations use Perplexity every day, asking more than 1.5 billion questions a month. IREN AI Cloud delivers the capacity and consistent performance to power AI workloads at scale.
IREN’s own post on 30 October says that under a multi-year agreement Perplexity will run workloads on NVIDIA Blackwell Ultra GPUs at an IREN data center campus, supporting the models and agents behind Perplexity Computer, and that Perplexity users ask more than 1.5 billion questions a month.
Sharing early evals on Gemini 4 Carbon. Strongest model that we’ve seen so far for long-form research and answer quality, with a new SoTA on the Wide-And-Deep-Research (WANDR) benchmark. Coming soon to all Perplexity Pro and Max users inside Computer, once we have sufficient token capacity. Congrats @GoogleDeepMind!
Perplexity CEO Aravind Srinivas posted on 30 October that Gemini 4 Carbon is coming soon to Perplexity Pro and Max users inside Computer, once Perplexity has sufficient token capacity.
$IREN just confirmed Perplexity Computer will run on $NVDA Blackwell Ultra GPUs at its data centers under a multi-year agreement.
Interesting considering AI agents use far more compute per task and IREN still has ~$3B in contracted ARR waiting to come online.
@StockSavvyShay’s post says IREN confirmed Perplexity Computer will run on Nvidia Blackwell Ultra GPUs under a multi-year agreement, and puts IREN’s contracted ARR still to come online at about $3 billion; that figure is the poster’s.
The Business Lever: Capacity for Agent Workloads
Earlier on 30 October, Perplexity chief executive Aravind Srinivas wrote that Gemini 4 Carbon is “Coming soon to all Perplexity Pro and Max users inside Computer, once we have sufficient token capacity”. We covered that post in an update to our Gemini 4 Carbon piece. IREN’s post names Perplexity Computer as the product its capacity will support.
On the supplier side, IREN’s 27 August release says “2026: capacity largely sold out” and lists “Recent 3-year contracts >$20m revenue per MW (IT)”. IREN defines ARR as GPU/hour pricing for commissioned GPUs under contract multiplied by 8,760 hours, and notes that it is “not a substitute for, GAAP revenue”.
The Structural Read
As we read the two IREN posts, the new detail is the product and the chip: IREN first listed Perplexity as a customer in July, and on 30 October it named Perplexity Computer and NVIDIA Blackwell Ultra GPUs.
As we read Srinivas’s post, Perplexity tied a model rollout inside Computer to having “sufficient token capacity” on the same day IREN named Computer as the workload its capacity will support.
As we read IREN’s August release, contracted run rate and operating run rate are different numbers: IREN says ARR ramps after commissioning, testing and customer acceptance in the months after each data center’s delivery.
Tim Delcourt, Commercial Director, AI Cloud at IREN, 30 October 2026
“At IREN, our job is to keep capacity ahead of AI ambition.”
Three Implications
FOR PERPLEXITY Nate Kupp says dedicated compute is one element of making outcomes more reliable and affordable, IREN’s blog quotes him as saying.
FOR IREN IREN’s August release says 2026 capacity is largely sold out and lists recent 3-year contracts above $20 million revenue per MW (IT).
FOR THE NUMBERS IREN gives no value, term or size for the Perplexity agreement, so the $3 billion gap is IREN-wide, as of 26 August, by our arithmetic.
The framework says “The structural bottleneck isn’t computing, it’s electrons.”
As we read it, IREN’s blog leads with the same input: it says IREN has 5GW of grid-connected power worldwide. The framework was published in December 2025 and does not discuss IREN or Perplexity.
What Is Not Established
IREN’s 30 October blog and X post do not give the contract’s value, term, megawatts, GPU count or campus location, and Perplexity had not posted about the agreement in what we read. The X post that prompted this piece, from @StockSavvyShay, says IREN has “~$3B in contracted ARR waiting to come online”; our $3 billion figure is the gap between IREN’s own two numbers as of 26 August, not a current figure.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
IREN says Perplexity will run workloads for Perplexity Computer on Blackwell Ultra GPUs at an IREN campus under a multi-year agreement, a customer it first listed in July. IREN’s own August figures put $4 billion of run rate under contract against $1 billion operating, a $3 billion gap by our arithmetic.
A note on sourcing. The agreement is described in IREN’s blog post and X post of 30 October 2026. IREN’s ARR figures come from its FY26 results release filed with the SEC on 27 August 2026; ARR is IREN’s operating metric, not GAAP revenue. The Srinivas quotation is from his post on X. We did not contact IREN or Perplexity. Figures labelled our arithmetic are our own.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
