Curtis Slipman

Founder of Penn’s Spine Center using first principles to analyze markets, energy, AI infra, uranium, and medicine.

[Curtis Slipman @slipmancurtis]

@GlobalCollapse @ZeekTyt Full DV apex is funded at 80% or less:

[MTS @MTSlive]

Microsoft would rather build its own AI data centers. So why is it relying on neoclouds?

Nebius CRO @marcboroditsky explains the 3 competing demands driving that decision:

"Microsoft and all the other hyperscalers would actually have preferred to build their own data centers like they have historically. It's a tough place that they all find themselves in."

"Copilot is massively successful, and they need the compute to do that. They're also committing to building their own models, and they're well behind their originally anticipated delivery. And on top of that, they have a cloud compute business."

"So they're trying to serve three masters."

"Our business is to actually sell to the AI natives and the enterprises out there that will be the ultimate customers for all of us."
@nebiusai

[Curtis Slipman @slipmancurtis]

Agree 💯. I would add that since enterprise and sovereign relationships become particularly sticky within 1.5 yrs to 2 yrs then the operating headline should be “time to stickyness” and not time to compute. IREN will be competing with other neoclouds and ought pin down those relationships asap

[Curtis Slipman @slipmancurtis]

In my view, IREN's Mirantis acquisition is IREN's read on where AI is moving. Enterprises and state entities increasingly will not hand over their data to hyperscalers. Cybersecurity exposure when your data sits on infrastructure targeted by nation-state actors is a vulnerability sophisticated organizations cannot accept after the Iran War. Regulatory compliance pressure builds as hyperscalers increasingly compete with their own customers. Adobe cannot have Microsoft seeing the customer images Photoshop AI trains on when Microsoft sells competing creative AI. The durable moat in an age where Claude Code commoditizes application layer software comes from infrastructure built around your specific use case, which only owners can build. That owned infrastructure also captures the long run cost advantage at scale plus escape from hyperscaler pricing power. State entities layer on top of all this because sovereign AI requires infrastructure controlled by the sovereign and not by a US hyperscaler subject to US data access requirements. Every country with serious AI ambitions needs this and most cannot build it themselves. Mirantis gives IREN bare metal at GW scale with orchestration software the customer controls allowing them to keep their own data and make their optimal operational decisions. Of course, this ultimately means that IREN will be competing directly with hyperscalers for the same customers and offering them what hyperscalers structurally cannot: ownership and control. This is the irrevocable path that will require a C-suite like IREN's to cross the finish line.

[Curtis Slipman @slipmancurtis]

@BitcoinAIGuy 💪💪Bullish💪💪

[Curtis Slipman @slipmancurtis]

$IREN is priced on disclosed revenue against a disclosed GPU fleet. The Lenovo partnership spans hardware, software, inference optimization, lifecycle support and early hardware access, funded at 3% dilution with first year payback. TurboQuant is the first published quantification of what that integrated structure is worth. $IREN

[Curtis Slipman @slipmancurtis]

CoreWeave captures roughly 70 to 80% of $IREN's benefit given strong CUDA engineering, but the absence of OEM co optimization slows production integration. Lambda reaches 40 to 50% with no structured optimization capability. Hyperscalers gain the largest absolute benefit and zero differentiation since all adopt simultaneously

[Curtis Slipman @slipmancurtis]

Do not ignore this signal. Single-site 750MW, owned power, mapping flexibility, cooling optionality and a single rather than fragmented multi-site campus (CoreWeave) places IREN at the head of the list. Why? The physics of Rubin demand exactly this architecture. Childress was built for this moment. $IREN”

[Curtis Slipman @slipmancurtis]

@FransBakker9812 @RemindMe_OfThis https://t.co/yYSDuCdU6v

[*Walter Bloomberg @DeItaone]

TRUMP TO ANNOUNCE DATA CENTER ENERGY DEALS DURING STATE OF THE UNION- POLITICO

[Curtis Slipman @slipmancurtis]

Sinclair at the World Government Summit saying he reversed biological age in animals by 50-75%. Let’s create a framework of facts to provide perspective on this fantastical claim. GSK bought his resveratrol company for $720M, Pfizer and Amgen couldn’t replicate the mechanism and GSK wrote off the whole $720M. Problem was with his methodology. Proper and reproducible methodology is basic for any lab no less one based at Harvard. OvaScience, a fertility company he co-founded went under as the science did not translate to expected clinical outcomes. His dog aging supplement led to his resignation from the Academy for Health and Lifespan Research after Matt Kaeberlein called it a lie. His 2023 paper claiming chemical cocktails reverse cellular aging was published in a journal where he serves as co-editor-in-chief and cleared peer review in 12 days — a process that typically takes 14 weeks. There is an FDA cleared Phase I vision trial that is worth following. But Sinclair on a stage making grand claims with a company waiting in the wings is not new information and must be viewed with skepticism given the fact pattern

[Curtis Slipman @slipmancurtis]

@AlemzadehC 🙏 thank you

[Curtis Slipman @slipmancurtis]

@AlemzadehC excellent analyses. thank you. when you have a moment could you comment on IREN, CCJ and EQX. Thank you