Monologue: We Need To Talk About CoreWeave - Better Offline

Summary

Podcast Summary: Better Offline – "Monologue: We Need To Talk About CoreWeave"

Episode Information

Title: Monologue: We Need To Talk About CoreWeave
Host: Ed Zitron, Tech Industry Veteran
Release Date: March 21, 2025
Podcast: Better Offline by Cool Zone Media and iHeartPodcasts

Introduction

In this episode of Better Offline, host Ed Zitron delves deep into the precarious financial and operational state of CoreWeave, an AI cloud provider. Zitron critically examines CoreWeave's business model, financial health, and its strategic relationships within the generative AI industry. This monologue serves as an incisive analysis of CoreWeave's viability and its broader implications for the AI infrastructure sector.

CoreWeave's Business Model and Financial Health

Ed Zitron begins by outlining CoreWeave's position in the AI cloud market, highlighting its focus on providing GPU compute resources to AI companies for running and training models. Despite its strategic role, CoreWeave's financial fundamentals appear shaky.

Revenue and Losses: Zitron points out that in 2024, CoreWeave generated just under $2 billion in revenue but incurred a staggering loss of $863 million. He emphasizes the unsustainable nature of this financial trajectory, stating, "If this company was in any other industry, it would be seen as a big pile of shit" ([04:15]).
Debt Burden: CoreWeave is heavily indebted, carrying approximately $8 billion in debt with high-interest rates. This debt structure forces the company to allocate upwards of $1.5 billion annually in loan payments, severely limiting its financial flexibility and growth potential.

Dependency on Major Clients: Microsoft and Nvidia

A significant concern Zitron raises is CoreWeave’s heavy reliance on a single client, Microsoft, which accounts for 62% of its revenue.

Microsoft's Pullback: Recent reports indicate that Microsoft is retracting from expanding its data center capacity and has reduced some contracts with CoreWeave. Although CoreWeave has denied these allegations, the uncertainty remains palpable. Zitron remarks, "62% of that revenue is from Microsoft, which has materially pulled back on datacenter buildouts and may have dropped some CoreWeave contracts" ([07:45]).
Nvidia’s Role: Additionally, Nvidia constitutes 15% of CoreWeave’s revenue. Nvidia not only invests in CoreWeave but also provides it with priority access to crucial GPU chips. Zitron muses, "With CoreWeave getting some products before even Amazon or Microsoft, it's bonkers" ([06:30]).

CoreWeave’s Expansion Plans and Partnership with Core Scientific

CoreWeave has ambitious plans to expand its compute capacity by 1.3 gigawatts, aiming to support future growth and service increasing demand from generative AI applications.

Partnership with Core Scientific: To achieve this expansion, CoreWeave has partnered with Core Scientific, a company infamous for its failed SPAC merger and subsequent bankruptcy in 2022. Core Scientific's expertise lies in Bitcoin mining, utilizing specialized ASIC chips that are fundamentally different from the GPUs required for AI compute. Zitron criticizes this partnership, noting, "Core Scientific... their data centers are not just a plug and play thing. You basically have to demolish them" ([08:50]).
Feasibility Concerns: The partnership raises red flags about the feasibility of CoreWeave’s expansion. Core Scientific has minimal experience in AI infrastructure, having generated only $24 million in 2024 from AI compute-related services, which is just 5% of their total revenue. Zitron underscores the improbability of successfully scaling AI compute capabilities through a partner with such limited relevant expertise.

Implications for the Generative AI Industry

CoreWeave's struggles are emblematic of broader challenges within the generative AI infrastructure sector.

Viability of Generative AI Businesses: Zitron argues that CoreWeave’s financial instability casts doubt on the profitability and sustainability of companies within the generative AI space. "CoreWeave makes its money by renting out GPUs to companies looking to run or train their models, which means that it's incredibly indicative of whether there's a real business in Generative AI. And you'll be shocked to hear that there isn't" ([05:20]).
Market Reality Distortion: Despite the high demand for generative AI, CoreWeave's financial woes suggest a discrepancy between market hype and actual business viability. Zitron states, "CoreWeave... exists within its own reality distortion field" ([05:50]).

IPO and Future Prospects

CoreWeave recently filed to go public with an anticipated valuation between $25 to $30 billion, marking it as one of the largest IPOs in the current generative AI boom. However, Zitron remains skeptical about the company's prospects.

IPO Challenges: The IPO is unlikely to provide sufficient capital to address CoreWeave’s extensive debt and funding needs. Zitron estimates that the company requires at least $30 billion to fund its expansion, far exceeding what the IPO can secure. He highlights, "CoreWeave... needs to invest tens of billion dollars that CoreWeave does not have" ([09:00]).
Sustainability Concerns: Given the company's financial deficits and dependency on a dwindling client base, Zitron questions whether CoreWeave can sustain its operations post-IPO. "CoreWeave had intended to go public a week ago... this is the big IPO. This is the big ai IPO. I am going insane every time I read about these goddamn companies" ([08:10]).

Conclusion

Ed Zitron wraps up his monologue with a stark assessment of CoreWeave’s future, suggesting that the company's financial mismanagement and strategic missteps could spell disaster, not only for itself but also for the broader generative AI industry. He emphasizes the need for investors and industry stakeholders to critically evaluate the sustainability of AI infrastructure companies like CoreWeave before getting swept up in the hype.

Final Thoughts: "If Generative AI had this incredible demand... CoreWeave would be making fat stacks of cash and have a far more diverse customer base... But it is what it is, I guess" ([09:50]).

Zitron concludes with a cautionary note, urging listeners to remain vigilant about the financial health and strategic direction of companies pivotal to the AI ecosystem.

Key Takeaways

CoreWeave faces significant financial instability, with high debt and heavy reliance on a single major client.
The partnership with Core Scientific raises serious doubts about the feasibility of CoreWeave’s expansion plans.
CoreWeave’s struggles reflect deeper issues within the generative AI infrastructure market, questioning the sector’s overall viability.
The impending IPO may not provide the necessary capital to stabilize CoreWeave, potentially leading to broader repercussions in the AI industry.

Notable Quotes

"If this company was in any other industry, it would be seen as a big pile of shit." — Ed Zitron ([04:15])
"With CoreWeave getting some products before even Amazon or Microsoft, it's bonkers." — Ed Zitron ([06:30])
"CoreWeave... exists within its own reality distortion field." — Ed Zitron ([05:50])
"CoreWeave would be making fat stacks of cash and have a far more diverse customer base... But it is what it is, I guess." — Ed Zitron ([09:50])

This comprehensive analysis by Ed Zitron provides listeners with a critical lens through which to view CoreWeave's role in the generative AI boom, highlighting the potential pitfalls of rapid expansion in a volatile market.

Transcript

Shannon Schuyler (0:00)

In a world of economic uncertainty and workplace transformation, learn to lead by example from visionary C Suite executives like Shannon Schuyler of PwC and Will Pearson of iHeartMedia, the Good Teacher explains the great.

Will Pearson (0:15)

Teacher inspires don't always leave your team to do the work that's been the most important part of how to lead by Example.

Shannon Schuyler (0:24)

Listen to Leading by Example executives making an impact on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

Ed Zetron (0:35)

Call Zone Media Bird Up. I'm Ed Zetron and this is your weekly Better Offline monologue. Better Offline so last week I put out a newsletter about CoreWeave, an AI cloud provider that sells GPU compute to AI companies looking to run or train their models, and they recently filed the paperwork to go public now. The newsletter explained the shaky fundamentals of the business, which is putting it mildly, and asked the question, how the fuck does coreweave survive? This is a company that's drowning in debt and reliant on one company for more than half of its revenue. And said company, Microsoft, is actively pulling away from building new data center capacity and has reportedly pulled back from some contracts with coreweave, though it's not clear what those contracts are. Now, this is an important story, and one that raises questions not just about the viability of Core Weave, but the generative AI industry at large. That said, the subject matter is a little dense to turn into a hulking two part episode, so I decided to condense it into a shorter monologue for you. Now, coreweave had intended to go public a week ago. They might go public in a week from now, and they're apparently going with like a 25 to $30 billion valuation. While it's hardly a recognizable name like say, OpenAI or Microsoft or Nvidia, this company is definite worth taking a look at, if not for the fact that it's arguably the first major IPO that we've seen from the current Generative AI hype bubble, and it's undoubtedly the biggest. Moreover, coreweave is a company that deals in the infrastructure aspect of AI, where one would naturally assume is where all the money really is, putting up servers for hyperscalers to run their hallucination prone unprofitable models. If Nvidia, the company that makes the GPUs for generative AI, is selling the pickaxes for the current Gold Rush corporate Core weave, well, they own the land. Or maybe they're the shovels. I'm not going to get too far into that metaphor as an Aside, up until a few years ago, Core Weave was a different company. A cryptocurrency mining data center company, to be specific. Now, the 2010s were turbulent for crypto, and the company then pivoted to providing high performance computing for third parties that didn't want to run their own infrastructure for things like 3D modeling. Around the time of the launch of ChatGPT, Core Weave pivoted again, this time towards providing compute to hyperscalers and then generative AI interest. It gambled and it gambled successfully. You'd imagine that such a company would be a thriving, healthy business though, right? They, I mean, they, they sell the surfer architecture. The demand is there, the growth is there. I mean, everyone wants generative AI, right? It wouldn't be another thing where the reality. Jesus Christ. Even a cursory glance at Corey's financial disclosure documents reveals a business that's precarious at best. And in my most uncharitable opinion, it's utterly rancid. If this company was in any other industry, it would be seen as a big pile of shit. Except it's one of the standard bearers of the generative AI boom and so exists within its own reality distortion field. Now, Core Weave makes its money by renting out GPUs to companies looking to run or train their models, which means that it's incredibly indicative of whether there's a real business in Generative Generative AI. And you'll be shocked to hear that there isn't. Core Weave made just under $2 billion in 2024, yet somehow managed to lose $863 million. Worse still, 77% of its revenue comes from two companies. Microsoft, 62%, and although not said explicitly, most likely Nvidia, 15%. Now the information is reported Project Osprey, which is basically Nvidia's stake and the fact that they're funneling money to them. It's actually not immediately obvious what it is that Nvidia is paying Core Weave for, though. Nvidia is both invested in the company and provides it with priority access to its chips. With Core Weave getting some products before even Amazon or Microsoft, it's bonkers. And again, you'd think with all this preferential accent, Core we've just printing money, right? Versus burning it. Burning it by the hundreds of millions full. I wish I could get this much money and just. I don't know. I would be doing a far weirder podcast. It would be exactly the same thing. Better offline would exist. I would just have hundreds of millions of dollars, we'd have just like Samuel L. Jackson coming in for some episodes. Anyway, I don't have that money. If you'd like to give it to me, please, please do. Now, a few weeks ago, the Financial Times reported that Microsoft had pulled out some of their Core Weave contracts, though Core Weave denied the allegations, and mysteriously, A week later, OpenAI announced that they had an $11.9 billion contract with Core Weave to provide compute services. Now, while I can't say for certain, can you think of another reason Microsoft needed a whole bunch of GPU compute, other than to host OpenAI's models? Hmm. Now, I'm just guessing here, but one has to wonder if it wasn't so much OpenAI signing a new deal and more IT taking over the future compute that Microsoft was formerly handling. And yes, you're remembering correctly. Microsoft has recently canceled more than the gigawatt of future compute capacity. Though from what I've heard, it's Core Weave contracts, future or otherwise, weren't included in that number. Anyway, for Core Weave to make that $11.9 billion in revenue from OpenAI, they're going to need compute, which will cost them tens of billions of dollars to build, which will be difficult as the company is burdened with about $8 billion of debt with horrifying interest rates, which may lead it to pay upwards of $1.5 billion a year in loan payments, with one of them requiring Core Weave to repay the loan with any other debt they raise in the future. Core Weave, by my estimates, needs at least $30 billion to expand. It has about $1.3 billion in the bank and just under $4 billion left on its loans that it can draw. One of them is a term loan where you can draw more up until about June 2025. Although Core Weave will raise capital in its IPO, the amount won't be anywhere near enough to meet its needs. And when I say its needs, I mean like the ability to service the revenue that they need to spend more money really does not make a lick of sense when you think about it. And you may wonder if things can get worse. And the answer is that they can. Core Weave has a planned 1.3 gigawatts of expansion, and its partner to build it is a public company called Core Scientific. Different company, similar name, genuinely different. Here are some facts about Core Scientific. They went public in 2022 in a disastrous SPAC merger and then filed for bankruptcy in the same year. Their entire business has been focused on mining Bitcoin, which requires specialized ASIC computer Chips that are entirely different, both in their construction and their maintenance to GPUs. And they can only really be used for mining crypto. They're not easily repurposed for other tasks. And indeed, their data centers are not just a plug and play thing. You basically have to demolish them. And another thing about Core Scientific, they only made $24 million in 2024 from selling AI compute related services to their customers. That's around 5% of its total revenue, with the rest coming from mining and selling crypto. Also, really dumb ass mistake I just make there. I said customers, I meant customer. Core Scientific has one customer, and that customer is. It's CoreWeave. As of right now, Core Weave has approximately 360 megawatts of compute power. Somehow it intends to build another 1.3 gigawatts worth of compute using a partner that has never built an AI data center and does not appear to have any meaningful compute right now. This is all so good. I love this. Let's fucking go. This is the big ipo. This is the big aiapo. I am going insane every time I read about these goddamn companies. But let's summarize. Coreweave is burdened by interest payments that may balloon to as much as $2 billion a year, and lost $863 million on $2 billion of revenue in 2024. 62% of that revenue is from Microsoft, which has materially pulled back on datacenter buildouts and may have dropped some Core Weave contracts, though coreweave denies this is the case. Of course, Corby's expansion, which is critical to servicing future revenue and growth, requires it to invest tens of billion dollars that Corby does not have. Corey's data center expansion is dependent on what is primarily a bitcoin mining company called Core Scientific that appears to have no AI capacity building. And they're meant to build over a gigawatt of capacity, a time where they do not appear to have done so. And as a reminder, converting cryptocurrency mining data centers to HPC data centers is effectively starting from scratch. It's not good. It's not good at all. Corby should have been a positive signal for Generative AI, or at least the way for Generative AI to kind of go further than where it is today. Or at least a way for AI boosters to shut me up. If Generative AI had this incredible demand, both from companies looking to integrate it and users looking to use it, coreweave would be making fat stacks of cash and have a far more diverse customer base and if I'm honest, not have to take up more than five times its revenue and burdens from loans with loan shark level interest rates just to survive. But it. It is what it is, I guess. And another note, Jim Cramer of CNBC. He said that Coreweave was going to be one of the biggest IPOs of the year. That should tell you about.