Luis Ureta

Mining Mafia

[Gavin Baker @GavinSBaker]

Market is overreacting to hyperscale credit spreads widening from my perspective. TL;DR Spot pricing for renting GPU compute materially above contracted rates implies hyperscalers are underearning while operating cash flow acceleration is an underestimated source of funds for AI capex.

The fact that spot prices for GPU rentals are at least 2x higher than contracted rates is the missing piece from the discussion about hyperscaler credit, which is the only fundamental factor behind this selloff. Multiple private companies are planning on spending at least 2x more per GPU for compute as contracts roll-off and some have spoken about this publicly.

As contracts roll-off, hyperscale growth rates are going to continue to accelerate as their installed bases of compute reprice higher. Hyperscale operating cash flow growth using a mix of estimates and actuals is modeled to accelerate from 31% in the first quarter of 2026 to 50% in the second quarter. This acceleration should continue for the rest of the year and this is not in estimates which incorrectly model a deceleration in the third quarter from my perspective.

Some math. Consensus estimates are probably for 25-35 gigawatts added by hyperscale and neoclouds in CY28 (using a range as standing up datacenters is hard and a lot of the neos plus labs are still private).  At 60b per gigawatt, that is 1.5 to 2.2 trillion in capex. Consensus estimates for hyperscale/neo operating cash flow is 1.3 to 1.4 trillion. I think this gets revised up materially as contracts reprice and growth accelerates so the 100b to 700b that would hypothetically need to be plugged by debt goes away. And their credit profiles materially improve. Not to mention the said 100b to 700b would be less than 1 turn of incremental leverage on consensus EBITDA estimates. And obviously the Nvidia and Broadcom “credit wrappers” help improve creditworthiness as well given their FCF profiles.

OpenAI, Cursor/Grok and the various Open Source inference clouds have accelerated materially over the last two months per public data and Anthropic continues to grow insanely fast while likely generating FCF. This - along with the fact that spot prices for GPU rentals are so far ahead of contract - are the missing pieces from the BofA chart on hyperscale FCF vs. semiconductor FCF.

Hyperscalers are underearning and anyone who signed a contract for GPU compute in 2024 and 2025 is overearning. Operating cash flow will be enough to fund capex but as contracts reprice and cloud growth continues to accelerate then spreads likely come in as well. 

Would also note that CDS markets are easy to manipulate - was a huge feature of the GFC - short the stock and then buy the CDS. So I would not put attach much signal to CDS.

Net, net I’m not that concerned about the widening spreads in hyperscale credit. The real risk is that bringing power online and energizing all these GPUs is really hard but we are getting better at this every day.

[Nebius @nebiusai]

Openness keeps AI competitive, adaptable, and available to everyone. That's why Nebius has joined @NVIDIA and 70+ companies in signing the open letter supporting open-weight AI models.
A thriving AI economy depends on choice, sound economics, and the freedom to deploy models wherever they're needed: https://t.co/AhMrs5mDXx

[Luis Ureta @luismo09]

@BlackPantherCap LFG

[Gavin Baker @GavinSBaker]

Nvidia is the leading open source AI company by a wide margin.

The silliness of this post is compounded by the fact that the letter did not suggest that closed source AI companies should open source their models.

[Ryan Hanrui Wang @hanrui_w]

The future of AI will not be built by one company, one country, or one model.

At Nebius Token Factory @nebiustf , we support open models because they give more builders the freedom to experiment, customize, deploy, and innovate.

Open and closed frontier models both have an important role to play. A thriving open ecosystem ensures that the benefits of AI can reach every industry, company, and country.

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Gallery St. 갤스 @J90236317]

버핏이 CNBC에서 구글 얘기한 2분이 요즘 AI capex 논쟁 전체를 정리함.

"구글이랑 경쟁자들 전부 수천억 달러를 깔고 있다. 그건 진짜 돈이다. 우리 철도 개발에도 그런 돈은 안 들어갔다. 소프트웨어 시절엔 없던 게임이다."

자산 가벼울 땐 안 사더니 capex 무거워서 주주들이 싫어하는 지금 샀냐는 질문엔 — "기록을 보면 이 회사들이 월스트리트가 파는 물건의 90~95%보다 승자가 될 확률이 높다. 월스트리트 보고서 중에 기업의 실제 내부수익률을 파고드는 걸 본 적이 없다. 다음 분기 얘기만 묻는다. 우스운 일이다."

버크셔는 이 판에 310억달러를 얹었음. 어제 구글 FCF가 22년 만에 처음 마이너스가 된 바로 그 지출을, 버핏은 사야 할 이유로 읽는 중.

[Nicolas Bustamante @nicbstme]

Guys, the world is so compute constrained. I honestly don’t understand how anyone can think otherwise.

Almost nobody is seriously using AI today, and we’re already hitting capacity. Everything is a bottleneck: land for data centers, permits, electricity, grid connections, chips, construction, cooling.

Demand is growing far faster than supply, and expanding supply is insanely expensive.

In this game, the best-positioned companies are the ones with massive net income from non-AI products. They can fund hundreds of billions in CapEx while everyone else has to raise money just to stay in the race… and remember if you are not at the frontier you have no pricing power so you won’t be able to fund your compute need.

[Chris Camillo @ChrisCamillo]

That didn’t take long

[Luis Ureta @luismo09]

@BitMNR Worst investment ever, thankfully sold this dog with fleece. Don’t care about the huge loss, I’ll sleep better knowing that this trash is no longer part of my investments. Nevertheless, wishing luck for everyone

[Melvin @MelvinInvests]

Nebius will be the first neocloud to hit $1 trillion and this morning's news just confirmed it (Save this).

Every neocloud on earth has the same constraint, you can only grow as fast as you can build.

CoreWeave raises billions in debt to buy GPUs, Applied Digital scrapes together capital for data centers, IREN fights for power contracts and the physical buildout is always the ceiling.

Nebius just removed their ceiling entirely.

This morning, Nebius announced an asset-light infrastructure partnership model where outside partners finance and own the physical infrastructure.

Nebius supplies the architecture, software stack, operational expertise and most importantly, the customers.

The partner builds the data center and buys the hardware, while Nebius provides the technology platform, trains the team, and sells the compute capacity.

The partner can generate returns immediately because Nebius brings the demand through its existing global customer base.

Nebius captures licensing fees, revenue-sharing, and commissions with zero additional capital required on their end.

This is not a neocloud model anymore but rather a franchise model for global AI cloud infrastructure and there is no comparable precedent in this space.

Nebius already had $46 billion in contracted revenue from Meta, Microsoft and a strategic 8.3% stake from Nvidia.thestreet

Today they opened the platform to every data center owner in the world.

When you model what happens if even a fraction of global data center investment starts flowing through the Nebius platform at software like margins, with almost no capital deployed, the upside becomes enormous.

The companies that reached a $1 trillion market cap did it by owning the platform everyone else built on top of and that is exactly what Nebius just became.

Extremely bullish on Nebius and make sure to follow me @melvininvests for more underrated gems.

[M. V. Cunha @mvcinvesting]

JUST IN:

$NBIS introduces business model to scale AI cloud globally through infrastructure partnerships 👀

Under the model, partners finance and own the infrastructure and hardware, and operate the data centers. Nebius supplies its systems architecture and supply-chain access, deploys and maintains its hardware design and software and services stack on the partner infrastructure, and takes the resulting capacity to market through its global sales organization.

Partners get fully-owned AI infrastructure assets, designed to Nebius standards, and a fast route to serve the AI cloud market. Nebius’s architecture and platform transform a partner’s raw capacity into a production-ready AI cloud, which Nebius then connects to customers. Because Nebius brings the demand, partners can begin generating a return as soon as the capacity goes live.

For Nebius, this asset-light approach expands the capacity it can offer its customers, such as AI natives and enterprises, with minimal incremental capital requirements. Partners’ data centers will join the Nebius capacity pool, adding incremental capacity to that coming online from Nebius’s owned data centers and colocations.

Arkady Volozh, Founder and CEO of $NBIS, said:

“Our new asset-light model gives infrastructure partners a flexible way to benefit from the explosive growth of AI. Our software allows partners to reach a much wider customer base with much better margins than conventional wholesale bare-metal contracts. We’re inviting data center investors, regional partners and others with capacity or capital to contribute to join us in serving this demand, combining their assets and local strengths with Nebius’ technology, platform, operational expertise and customer demand.”

Nebius anticipates pursuing a variety of economic arrangements under this partnership model, including revenue-sharing agreements, licensing fees and commissions, as well as committed capacity arrangements that would provide Nebius with access to additional compute to be sold to customers. The company has already entered into initial arrangements under this asset-light model.

As part of the partnership agreements, Nebius will equip partner teams to run the site and will remain responsible for the cloud software and service levels, while the partner manages the facility and hardware. Customers receive the same standard of service whether they run on Nebius’ own infrastructure or a partner’s.

[Luis Ureta @luismo09]

@AndreasSteno great call

[M. V. Cunha @mvcinvesting]

🚨 $NBIS is preparing its first data center in India.

Just two hours ago, the company posted two on-site job openings in Hyderabad, both related to new data center launches.

Asia, here we go.

I first discussed this possibility in June last year, after The Economic Times reported that India-based data center companies were looking to attract neocloud firms to the country.

h/t @David917S for flagging this.

[Beth Kindig @Beth_Kindig]

Meta $META is expanding its Hyperion data center in Louisiana to 5GW, taking its planned investment in the facility to over $50 billion.

$NVDA $AVGO $AMD

[Nebius @nebiusai]

Nebius has taken delivery of NVIDIA’s Spectrum-6 102.4T Ethernet switch, part of NVIDIA’s latest Spectrum-X Ethernet networking platform. It's already in the lab at our Finland data center. Early access through our co-engineering with NVIDIA means the stack is verified before customer workloads ever run on it.

[TheBigBerbowski @TheBigBerbowski]

https://t.co/KGquxS2pzY

[Andreas Steno Larsen @AndreasSteno]

Sounds like a company with excess compute capacity to me

[Shay Boloor @StockSavvyShay]

$NBIS is expanding into Spain securing an 18MW lease with Merlin Properties at its Getafe campus.

The move marks Nebius first entry into the Spanish market and another step in scaling its European AI infrastructure footprint.

[Luis Ureta @luismo09]

@babyfolio 0