Why I take @danroberts0101 of $IREN seriously when he talks about building AI infrastructure at scale.
Before founding IREN with his brother Will in 2018, Dan had built a career financing, investing in and overseeing major infrastructure.
Airports. Wind farms. Solar farms. Gas pipelines.
Assets requiring substantial capital, technical expertise and reliable operations.
That background matters when your business involves turning land, power and billions of dollars into functioning AI infrastructure.
Start with Macquarie.
Dan spent six years across M&A advisory, debt structuring and principal investment, working in infrastructure, real assets and renewable energy.
Experience directly relevant to the decisions IREN faces:
How should projects be financed?
Where should capital be committed?
How should risks be allocated between investors, lenders, contractors and customers?
He also holds a Master of Finance from INSEAD, where he made the Dean’s List.
Then came Palisade.
Dan became an executive director and the second-largest individual shareholder of Palisade Investment Partners.
A 2020 account described the business as managing A$8bn of infrastructure across 23 businesses.
His responsibilities included originating opportunities, business development and overseeing portfolio assets.
That was Palisade’s portfolio value—not construction personally delivered by Dan. His experience was in making major infrastructure investment decisions and overseeing assets.
His board roles included Northern Territory Airports, Sunshine Coast Airport, Granville Harbour Wind Farm, Ross River Solar Farm, ANZ Terminals and the Tasmanian Gas Pipeline.
That is a substantial foundation in infrastructure leadership.
Now look at the team around him.
Michael Nudelman joined as Chief Development Officer with more than two decades of experience and senior roles at Google, CyrusOne and Beale Infrastructure.
His remit includes global data-centre development, site planning, permitting and economic development incentives.
John Gross, Chief Innovation Officer, brings specialist expertise in data-centre engineering, high-density infrastructure and liquid cooling.
His work on liquid-cooled servers dates back years, well before the current AI investment cycle.
Dan brings infrastructure investment and oversight experience. The team adds specialist development and engineering capability.
There is delivery behind the ambition.
IREN reported in August that Horizon 1—a 50MW liquid-cooled deployment at Childress—had been delivered to Microsoft.
That moves the story from securing land and announcing capacity to delivering infrastructure for a major customer.
The $9.7bn Microsoft agreement provides a substantial commercial foundation.
Mirantis adds software and cloud-services capabilities.
The NVIDIA relationship connects IREN’s infrastructure ambitions with the ecosystem supplying the compute.
IREN is bringing the pieces together:
Power → development → cooling → GPUs → networking → software → operations.
A co-CEO overseeing this expansion needs to recruit specialists, allocate capital, structure agreements and establish accountability across an enormous programme of work.
Dan’s career provides relevant experience for those responsibilities.
For me, the bullish case is the combination:
An experienced infrastructure investor leading the company.
Specialists with major industry credentials.
Large customers committing to its services.
Physical delivery demonstrating progress.
From two brothers founding an Australian company to delivering AI infrastructure for Microsoft.
Dan’s background helps explain how IREN reached this point—and why I’m excited to see what the team builds next.
$IREN
Sources below:
The @SemiAnalysis_ team on why Anthropic's S-1 is good news for neoclouds:
"Anthropic has shown today AI inference is a really great business. Even running the training is extremely profitable."
"Compute margins are excellent."
"Neoclouds are still very good businesses."
Revenue per GPU can increase as token prices decline if per GPU token throughput increases enough.
So both Anthropic revenue *and* margins can go up as token prices go down *if* GPU prices go up less than the product of (token price decline * per GPU throughput increase).
But I have super enjoyed all the commentary to the contrary!
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: https://t.co/TZSugXi2ZE
I've been calling out the importance of US Frontier and Open Source Models for Cyber Security. @BenBajarin provides a concrete datapoint here.
For Qwen3.8-Flash-Next, when "asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing".
We will see opportunity and investment in sovereign models for cyber security.
🚨BREAKING NEWS🚨 10/24 footage confirms massive expansion is underway at $IREN's Mackenzie site in BC.
Will @danroberts0101 discuss this on the 11/5 earnings call?
More details to come!
The sum of Anthropic, OpenAI, Gemini, Muse, Grok and open-weight revenue are what matter. Not any single one of those.
The better everyone ex Anthropic and OpenAI perform, the better for AI infra imho.
Noteworthy to me that open-weight models have not really ceded any share in the face of this Anthropic acceleration; i.e. the entire market is strong.
The notion that GPU debt and data center financing is *closed* is laughable.
As of now, the world is separated between those who say they know, and those who actually know.
If @jvisserlabs is right, intel is more valuable than equity, so let me tell you this:
selective credit markets means the due diligence goes beyond your traditional ways of finding out the winners from the losers.
I will gladly go on the record here and say that based on my latest, the debt markets are not closed for $IREN.
Best of luck to the others.
Swap lines activated:
*NVIDIA, AMD, BROADCOM INDICATE INTEREST TO BUY UP TO $5B OF META STOCK
Meta to Raise $35B in common, preferred stock; will launch separate ~$60B ATM no earlier than Q1