THE WORLD OF
Wes
Cummins
AI has a physical foundation
Power, cooling, land and the people who build. Wes explains the decisions that turn a computing ambition into infrastructure that can carry it.
Wes Cummins
Co-founder, Chairman & CEO · Applied Digital
Building infrastructure for AIPage settings
WES CUMMINS / ON MARKETSCALE
Start here.
Selected appearances.
Explore the work, the ideas and the people.
Architects of Acceleration
With Philbert Shih
As AI adoption accelerates at an unprecedented pace—ChatGPT alone sees 2.5 billion daily prompts just two and a half years after launch— digital infrastructure is racing to keep up. At the center of this transformation are purpose-built data centers, evolving from air-cooled Bitcoin facilities to liquid-cooled “AI factories” designed to power the next generation of computing.
Notes & credits +
As AI adoption accelerates at an unprecedented pace—ChatGPT alone sees 2.5 billion daily prompts just two and a half years after launch— digital infrastructure is racing to keep up. At the center of this transformation are purpose-built data centers, evolving from air-cooled Bitcoin facilities to liquid-cooled “AI factories” designed to power the next generation of computing.
Published articlePhilbert Shih · Host / Jason Zhang · Co-founder
Original on MarketScale ↗datacenterHawk
With David Liggitt
The company is racing to build out physical infrastructure capable of supporting the massive power and proximity requirements that modern AI compute demands
Notes & credits +
The company is racing to build out physical infrastructure capable of supporting the massive power and proximity requirements that modern AI compute demands
Published articleDavid Liggitt · Interviewer
Original on MarketScale ↗datacenterHawk
With David Liggitt
The speed requirements for AI shift dramatically depending on whether systems are learning or performing live tasks
Notes & credits +
The speed requirements for AI shift dramatically depending on whether systems are learning or performing live tasks
Published articleDavid Liggitt · Interviewer
Original on MarketScale ↗datacenterHawk
With David Liggitt
Building AI infrastructure where power is cheapest and most abundant eliminates transmission waste and cuts computing costs dramatically
Notes & credits +
Building AI infrastructure where power is cheapest and most abundant eliminates transmission waste and cuts computing costs dramatically
Published articleDavid Liggitt · Interviewer
Original on MarketScale ↗WES / FROM THE CONVERSATIONS
Ideas & insights.
Read the ideas, then watch, listen to or read the original work.
A different workload changes the building.
Wes Cummins and Jason Zhang with Philbert Shih
Wes and Jason Zhang describe moving from air-cooled Bitcoin facilities toward purpose-built, liquid-cooled AI infrastructure. Their conversation with Philbert Shih explores a clean-slate design and a repeatable modular approach. Experience from the earlier business informed the transition, while new hardware changed the requirements.
Discussion prompt
Which assumptions about power, cooling and layout need to be revisited when the workload changes?
Connected GPUs change the scale of the site.
Wes Cummins with David Liggitt
Wes tells David Liggitt how demand for closely connected GPUs changed Applied Digital’s real-estate planning. Power density and network proximity made scattered capacity less useful for some deployments. In the discussion, he describes shifting from smaller sites toward much larger facilities.
Discussion prompt
Assess the combined power and network needs of the deployment, rather than capacity alone.
Specify the latency of the actual task.
Wes Cummins with David Liggitt
Wes distinguishes long-running AI training from time-sensitive inference, batch generation and interactive applications. They do not all need the same response time or location. His discussion with David Liggitt starts with what a workload must do before deciding where it should run.
Discussion prompt
Does this workload require an immediate response, or can it operate farther from the user?
Location can begin with available power.
Wes Cummins with David Liggitt on stranded power
In his discussion of stranded power, Wes explains why some computing facilities can be placed near underused energy sources. The opportunity depends on the workload’s tolerance for distance and latency. It is a site-selection argument built around the relationship between the task and its physical requirements.
Discussion prompt
Evaluate the workload’s location constraints alongside the availability of power.
Efficiency connects operating costs with infrastructure.
Wes Cummins with Alicia Shapiro
With Alicia Shapiro, Wes discusses power usage effectiveness and the relationship between useful computing and the energy supporting it. He cites the company’s initial build as an example in that historical conversation. The discussion makes an efficiency metric part of the operating-design question.
Discussion prompt
Ask what a reported efficiency measure includes and which facility conditions it describes.
Execution depends on the team behind the capacity.
Wes Cummins on Applied Digital’s early growth
Wes attributes Applied Digital’s early expansion to recruiting experienced people and organizing around delivery. The short excerpt gives the human side of a conversation often dominated by megawatts and hardware. Capacity is an outcome of coordinated work, not only a number on a plan.
Discussion prompt
Who has the experience and responsibility to deliver each part of the plan?
Sources & how this brief was made
AI-assisted editorial summaries, reviewed against the linked published articles. Guest views stay attributed to the guest. Recurring themes and practical questions are editorial synthesis, not quotations or a statement of the host’s beliefs.
SHARED CONVERSATIONS
In conversation with.
THE RECORD
Career & credits.
13 verified contributions from MarketScale publications. Original titles and dates are retained; excerpts are not separate full episodes.
BACKGROUND & EXPERIENCE
About Wes.
Co-founder, Chairman & CEO
Applied Digital
Connect on LinkedIn ↗Wes Cummins is co-founder, chairman and CEO of Applied Digital. His career spans technology investing and digital infrastructure: he led technology investing at Nokomis Capital and founded 272 Capital, an investment firm focused on technology companies. He studied at Washington University in St. Louis. That investment background informs his work on the power, land and computing infrastructure required for large-scale AI facilities.
His MarketScale conversations explain the company’s shift from Bitcoin-focused facilities toward high-performance computing. With co-founder Jason Zhang and industry interviewers, he examines the physical choices behind that change: access to power, GPU proximity, cooling and the scale of a site. The catalog preserves the dates of these conversations so earlier plans and forecasts are not mistaken for current operating results.
Biography & career sources
About the media selection
13 verified contributions from MarketScale publications. Original titles and dates are retained; excerpts are not separate full episodes.
Summaries identify whether they use published articles or transcripts. Series attribution is distinguished from an individually checked appearance.
FROM THE PUBLICATION
Keep the conversation going.
Related reading from MarketScale’s publication, selected for the topics on this page.
Explore the publication ↗MARKETSCALE ·
As AI Chips Near 4,000 Watts, Liquid Cooling Becomes the Backbone of Data Center Design ↗
Surging demand and megawatt-scale rack roadmaps are pushing data centers from air to liquid cooling.
Read the article ↗Selected by MarketScale with AI assistance. Independent publication coverage; not attributed to or endorsed by Wes.Keep the conversation going.
Connect with Wes on LinkedIn.
Connect on LinkedIn ↗