Section Insights
Nvidia's Earnings Call Insights
What key insights can be drawn from Nvidia's recent earnings call?
Nvidia's earnings call revealed record revenues and significant growth in AI spending, indicating a robust future for the company and the AI sector.
- Nvidia reported record revenues of $96.2 billion, up 106% year-over-year.
- Operating income surged by 124%, with a 66.2% operating margin, outperforming competitors like Alphabet and AMD.
- AI spending is accelerating, suggesting a strong market demand for Nvidia's products.
Nvidia's Strategic Partnerships
How do Nvidia's partnerships impact its market position?
Nvidia's partnerships with major AI buyers and cloud providers position it favorably in the AI market, reducing risk and enhancing growth potential.
- Nvidia has contracts with major companies like Microsoft and Meta, ensuring a diverse customer base.
- The company is committed to purchasing unused compute capacity from partners, solidifying its role in the AI ecosystem.
- The AI cloud market is expected to triple in size, benefiting Nvidia's growth.
CoreWeave's Revenue and Growth
What makes CoreWeave a significant player in the AI infrastructure market?
CoreWeave's impressive revenue growth and substantial backlog of contracts highlight its critical role in supporting AI workloads and its strong relationship with Nvidia.
- CoreWeave reported $2.6 billion in revenue, up 112% year-over-year, with a backlog of $104 billion.
- The company has a formal partnership with Nvidia to purchase unsold cloud capacity, enhancing its operational stability.
- CoreWeave's growth is driven by contracts with major players like OpenAI and Meta.
Nebius's Unique Business Model
What distinguishes Nebius in the AI infrastructure landscape?
Nebius stands out due to its rapid revenue growth and a business model that relies on customer prepayments, reducing financial risk.
- Nebius achieved a 454% year-over-year revenue increase, with significant prepayments covering equipment costs.
- The company has $37.5 billion in contracted work, including major deals with Microsoft and Meta.
- Nebius's financial structure allows it to minimize debt and interest expenses compared to competitors.
Comparative Analysis of AI Infrastructure Companies
How do CoreWeave, Nebius, and Iren compare in terms of financial health and market position?
A comparative analysis reveals that while CoreWeave leads in scale, Nebius shows stronger growth potential with less debt and a more favorable funding model.
- CoreWeave has significant net debt, making enterprise value a crucial metric for evaluation.
- Nebius's higher market cap despite lower revenue run rate indicates strong investor confidence.
- The contrasting financial structures of these companies highlight different strategies for growth and sustainability in the AI market.
Transcript
0:00 Nvidia just had their best earnings call ever, but not for the reasons that you might think. Whether you're a long-time shareholder or you're just trying to figure out if we're in an AI bubble, you should know that this earnings call wasn't just about revenues or profit margins. Nvidia showed us exactly where the AI revolution is headed next and which stocks will win big as a result. My name is Alex and I spent eight years as an electrical engineer and AI researcher at MIT, which helped me find great stocks like Nvidia, Micron, and TSMC years before the rest of the market. So, let me show you where AI is headed next and how I'm investing in it to get rich without getting lucky. Your time is valuable, so let's get right into it. Nvidia reported earnings just a few days ago on August 26th and since they sit at the very center of the AI revolution, their earnings reveal a lot more than how much money they made. It tells us where that money is coming from, how fast it's moving, and where it's headed next. So, let's start with that. Nvidia reported record revenues of 96.2 billion dollars for the quarter, which is up 18% quarter-over-quarter and 106% year-over-year. So, right off the bat, AI spending isn't slowing down.
1:16 It's accelerating. Nvidia's year-over-year revenue growth was 62%, 73%, 85%, and now 106% over the last four quarters. Even better, their operating expenses only grew by 55% or around half as fast as their revenues. As a result, Nvidia's operating income shot up 124% giving them a whopping 66.2% operating margin. For context, Alphabet's operating margins are 34% and AMD's are around 17%. So, said another way, for every dollar they all make, Nvidia keeps twice as much as Google and almost four times as much as AMD, which are two of their biggest competitors in the AI chip market. There are lots of interesting things in Nvidia's earnings that probably warrant separate videos, like the fact that their earnings per share more than doubled, but about 15% of that growth actually came from $7.8 billion in gains on Nvidia's investments into other AI companies, which is up from $2.2 billion a year ago. Or the fact that they more than doubled their supply capacity commitments from $119 billion last quarter to $279 billion today, primarily to lock in memory for Vera Rubin this year and Rubin Ultra in 2027. That tells me that memory stocks will keep selling every single chip that they make over the next few years. So, the main metric to watch for memory stocks is how much they can increase their production capacity. Let me know in the comments if you want me to make a full video diving into Nvidia's earnings and what they tell us about every downstream AI market. But, the big thing I want to focus on today is one specific change to how they report their data center revenues. Last quarter, Nvidia stopped reporting data center revenues as a single number and started reporting them as hyperscale versus ACIE, which stands for AI clouds, industrial and enterprise. We all know the hyperscale companies, Microsoft, Amazon, Google and Meta Platforms. That means that ACIE includes everyone else from healthcare and financial services to automotive and robotics companies and from government and sovereign AI build-outs to AI cloud providers. There are two important facts about this category. First, ACIE grew by 138% year-over-year, while hyperscale grew by 102%. This is the first time that hyperscalers weren't the fastest growing segment, which means that AI build-out is broadening well beyond the trillion-dollar tech giants. And that broader spend is growing even faster.
3:58 And second, this category includes the neo clouds, specialized AI infrastructure providers like Coreweave, Nebius, and Airen, which are the focus of this video. One of them is the pretty clear winner, and I'll show you which one when we compare them. There are three big reasons to focus on neo clouds above other kinds of companies, at least for the near term. First, they have deals with every kind of AI buyer, from startups like Perplexity and Figure AI, all the way to the hyperscalers themselves. That means that even their direct competitors are their customers.
4:33 Microsoft alone has multi-billion-dollar contracts with all three neo clouds. And Meta Platforms signed a $21 billion contract with Coreweave and has a contract with Nebius that's worth up to $27 billion. Second, Nvidia itself is directly partnered with all three of these companies. Over the next five years, Nvidia is on the hook to buy any unused compute capacity from Coreweave and use Airen's infrastructure for their own internal workloads under a $3.4 billion contract. On top of that, Nvidia can buy up to $2.1 billion worth of Airen stock as they deliver GPUs. And they've already invested $2 billion each into Coreweave and Nebius. Investing in companies backed by the biggest player in AI and the most valuable company on the planet is a great way to lower your overall risk and get rich without getting lucky. And third, this entire category is on track to almost triple in size. Nvidia expects its neo cloud partners to finish 2026 with 8 gigawatts of installed capacity, up from roughly 3 gigawatts at the end of 2025. This is actually a virtuous cycle because more compute capacity means more companies and consumers can use AI more often and for more use cases, which means more demand, which causes the AI build-out to grow even faster. When people ask me which AI tool I open first, they're surprised when I don't say ChatGPT or a Claude. They're great, but the one that actually helps me the most is WhisperFlow, the sponsor of this video.
6:06 Most people talk way faster than they can type, especially with their thumbs. WhisperFlow is a super fast, high-quality voice-to-text tool that acts kind of like a keyboard replacement. You just hit the dictate button, talk to it, and it instantly transcribes everything you've said and puts it wherever your cursor is. But there are a few things that really set WhisperFlow apart for me. First, it gets everything right, even when I whisper. Second, it takes my messy speech, strips out all the 's and corrections, and turns it into polished, ready-to-send text. And third, it will remember the name of your business, your friends and family, and any other custom words that Siri or other built-in dictation tools always mess up for you.
6:50 You can even have multiple profiles with different settings to be more casual in your texts, but more formal with your punctuations and capitalizations in emails. Having all of that built into one tool that works across every app is super helpful. And if you haven't tried voice dictation yet, you're really missing out. WhisperFlow is free to try, but you can get 1 month of Pro with unlimited words by using my link in the description. All right, so CoreWeave, Nebius, and Iron are worth covering first because they're expected to triple their capacity over the next year.
7:21 They're all directly partnered with Nvidia, and even their competitors are their customers. Let's start with CoreWeave, ticker symbol CRWV, since it's the biggest of the three. CoreWeave operates 51 data centers purpose-built for AI across North America and Europe. They have 1.5 gigawatts of active power and 4.2 gigawatts of power under contract. We care about both numbers when we compare the Neo Clouds because they mean different things. Active power is the electricity actually flowing into servers that are running today, while contracted power is the promise of future electricity to a site that may not even be built yet. Only active power actually generates revenue. Nvidia's Blackwell racks hold 72 GPUs each, and they take around 120 kilowatts to power.
8:12 That means CoreWeave has enough contracted capacity to power over 30,000 racks, or more than 2.2 million Blackwell GPUs. Then they sell access to those GPUs either as bare metal chips or full stack instances that are managed through a cloud control plane. That lets CoreWeave spin up clusters of GPUs, add in low latency networking and storage, and schedule large AI workloads, all without touching the physical servers. On their latest earnings call, CoreWeave reported record revenues of $2.6 billion for the quarter, which was up 112% year over year. But like I've been saying for over a year now, the special thing about CoreWeave isn't their revenue growth.
8:54 It's their revenue backlog, which just reached $104 billion. That's up 246% year over year. That growth is mostly driven by multi-year, multi-billion dollar contracts with massive companies like OpenAI and Nvidia, Microsoft, and Meta Platforms. In fact, Meta alone signed another $21 billion deal with CoreWeave back in March. Another special thing about CoreWeave is their privileged relationship with Nvidia. CoreWeave is basically the launchpad for Nvidia's latest chips. Last quarter, they were the first to validate Nvidia's Vera Rubin racks, including the BlueField 4 GPUs and next-generation networking solutions that help make sure that all these expensive GPUs are being fully utilized. And like I said earlier, they have a formal $6.3 billion partnership where Nvidia will purchase any unsold cloud capacity from Coreweave through April of 2032. Earlier this year, Nvidia and Coreweave also agreed to build out more than 5 gigawatts of AI factories by 2030, enough to power millions more Nvidia GPUs. As part of this deal, Nvidia invested another $2 billion into Coreweave stock at $87 per share. That's actually higher than the stock trades today. So, anyone buying it now is getting a better price than Jensen Huang himself. Speaking of prices, there's one insanely important number that determines whether a neo cloud survives or thrives, and that's the price they pay on the money that they borrow to build their infrastructure in the first place.
10:28 Building AI data centers means paying for everything up front, the GPUs, the racks, the cooling, the power, all that gets bought years before they start making revenue. For example, Coreweave spent $9.4 billion on capital expenditures last quarter, even though they only made $2.6 billion. That's not a bad thing, it's the entire business model. Build it and they will come, but that means they need to borrow money to build it in the first place. And that means we need to look at metrics like their net debt, which is just their total debt minus their cash on hand. Net debt is what they would still owe if they emptied their bank accounts tomorrow. We also care about how much they pay in interest as a percentage of their total revenues.
11:11 Coreweave has about $35 billion worth of loans, another 16 billion in lease obligations for the data centers they don't own, and $5.5 billion in cash. 35 + 16 - 5 is roughly $46 billion in net debt, but they paid $640 million dollars interest last quarter alone, which is about 25% of the revenues they brought in. That makes a huge difference to their bottom line. Coreweave's operating losses for the quarter were 49 million dollars, but their net loss was 626 million. Almost the entire difference is interest. That's why net debt and interest expenses are important numbers to compare across all three companies.
11:56 Either way, the takeaway for investors is pretty simple. Nvidia is Coreweave's core infrastructure supplier, their launch platform, and their buyer of last resort, which seriously strengthens Coreweave's position in the neo cloud market. But what Nvidia can't do is lower Coreweave's interest expenses. All right, let's talk about Nebius next, ticker symbol NBIS, since they have the strongest balance sheet and the fastest growth in the group. Nebius is one of the most technically advanced neo clouds out there. They've already received their first Vera Rubin NVL72 systems, and Nvidia's CFO said that Nebius will be the first company anywhere to get the new Groq 3 LPX inference chips in volume. Nebius runs both kinds of AI workloads under one unified platform.
12:42 Aether is their cloud, where customers can rent GPUs and train their models, while Token Factory sits on top of it and handles inference. Nebius's inference workloads more than tripled last quarter alone, and more and more of that is coming from agentic AI, where a single task can trigger dozens of different tool and model calls. That means that revenue scales with how complex the task is, not just how many users sign up. And on the infrastructure side, Nebius just raised their contracted power target from 4 gigawatts to 5 by the end of 2026. Remember, that's contracted power, not active.
13:19 They expect to have 800 megawatts to 1 gigawatt of active power by the end of the year, and they're selling every single watt. Nebius has $37.5 billion worth of contracted work on their books, including a $17.4 billion deal with Microsoft and a contract with Meta worth up to $27 billion. The special thing about Nebius is that they get paid before they build. Roughly 70% of the deals they closed last quarter came with customer prepayments that cover 50 to 60% of the equipment costs, and they expect over $9 billion in prepayments this year. So customers are funding their AI buildout instead of high interest loans. Last quarter Nebius generated $582 million in revenue, which is up a whopping 454% year over year. And they're guiding for a $7 to $9 billion annualized run rate by the end of 2026, which would be close to a 7x from last year. Nebius has $8.5 billion worth of debt, around $1.5 billion in lease obligations, and $8 billion in cash. That works out to about $2 billion in net debt compared to Coreweave's $46 billion. Nebius paid $119 million in interest last quarter, or about 20% of their revenue. But these two companies are operating on very different scales. Nebius is doing under $600 million a quarter, but promising 5 GW of contracted power. As a result, they'll be spending $20 to $25 billion to get there this year. And part of that funding is selling stocks. So Nebius might be saving on interest, but their shareholders will pay in terms of dilution. And that leaves us with IREN, ticker symbol IREN, the former Bitcoin miner becoming an AI cloud. Last quarter IREN's AI cloud revenue hit $70.5 million and passed their revenues from Bitcoin mining for the first time. That AI number is up from $33.6 million dollars just one quarter ago, meaning it more than doubled in the last 90 days. What separates Iren from the other neo clouds is they already did the hard part. They spent years buying land and power contracts to mine Bitcoin, which means they already have a pipeline of more than 5 gigawatts lined up for their AI data centers. And that's already won them two enormous contracts. A 5-year 9.7 billion dollar deal with Microsoft and a separate 3.4 billion dollar deal with Nvidia.
15:51 Iren currently has about 7.8 billion dollars of debt and 5.9 billion in cash. So, about 1.9 billion dollars of net debt. They paid 25 million dollars in interest last quarter, or about 18% of their revenues. They also have a 16.6 billion dollar backlog and they're targeting a 4 billion dollar annualized run rate from this year's capacity. About a billion of that is already online today, which means they expect to quadruple it in the next 4 months.
16:22 Although their management did point out that the revenue they'll actually recognize could come in lower than that. And as usual, Iren posted a huge loss of 684 million dollars last quarter, most of which was a non-cash impairment for writing off more of their Bitcoin hardware as they convert their sites to AI. That's the ongoing cost of pivoting from mining to training and inference. And their total revenue actually shrank from 145 million dollars down to 137 million for the same reason. It's faster to take Bitcoin servers down than it is to stand AI servers up. All right, before we can decide which of these stocks is the best buy right now, let's compare them apples to apples. And if you feel I've earned it, consider hitting the like button and subscribing to the channel. That really helps me out and it tells me to make more content like this. Thanks. Now, let's compare Coreweave, Nebius, and Iren stock.
17:18 Here's a table summarizing everything I've covered. Keep in mind that I built it myself by pulling numbers from each company's latest earnings, and I tried to make every row as fair as I could. But, all three companies have different fiscal calendars, different contract lengths, and they're scaling from very different starting points from the beginning of the year. So, take this table as a good way to compare these companies, but not as official audited numbers. All right, here's what jumps out at me. First, Coreweave has a lot more net debt than the other two companies. So, we want to compare them in terms of enterprise value instead of market cap. When you buy a business, you buy its assets and its debts minus whatever cash they're holding. So, enterprise value is a much better way to price debt-heavy companies. Since Coreweave's debt is as big as their market cap, their enterprise value ends up being twice as big. That said, Coreweave is the clear leader in terms of scale. They're the biggest enterprise with the most revenue, the deepest backlog, and 1.5 gigawatts of active power actually running today. But, a quarter of every dollar they make goes right back to interest on their loans.
18:27 Nebius is the opposite. 454% revenue growth with only $2 billion of net debt, and customers pre-funding half of their entire buildout. What surprised me is that Nebius has a higher market cap than Coreweave with half the target revenue run rate and 1/3 of the backlog. That means Nebius' enterprise value is 7.4 times their year-end run rate against 4.9 for Coreweave and 4 for IREN. IREN is the smallest and cheapest company at a $16 billion enterprise value, but they also have the lowest revenue run rate. 500 million total and 280 million if you only include AI and not Bitcoin.
19:09 That's partially why they have the largest AI revenue growth of more than 10 X, but from a much smaller base. Which one is actually the better buy also depends on whether you care more about active power or contracted power. If you go by active power, CoreWeave costs around $62 million per megawatt, while Nebius and IREN are over $350 million each. So, CoreWeave is the cheapest by a factor of six, but if you go by total contracted power, IREN costs around $3 million per megawatt. Nebius costs $12 million, and CoreWeave costs $22 million. That makes IREN seven times cheaper than CoreWeave by contracted power. So, if you're the kind of investor that wants exposure to this category of stocks with the lowest execution risk, CoreWeave is probably the stock for you. They're the biggest company by far with the most active power being used for AI right now, and they have NVIDIA committed to buying any capacity they can't sell through April of 2032. If I was a newer investor or closer to retirement, CoreWeave is the one I'd pick. But if you're looking for the most upside, IREN is the smallest company with the most contracted power per dollar, the lowest interest payments relative to their revenues, and they're the cheapest based on their size and their target revenue run rate. The biggest risk for IREN is whether they can actually turn their contracted power into real revenues or if their pivot from Bitcoin mining will cost them too much too soon. I think this risk is already priced in. So, IREN is the one I'd go with if I could only pick one of these companies. And if you're not sure, Nebius sits right in the middle of them on almost every single metric. So, they might be the right stock for you as long as you're all right with them being much more expensive. Let me know which stock you're buying in the comments or if you want me to make a deep dive video on any one of them. And if you want to see what other stocks I'm buying to get rich without getting lucky, check out this video next. Either way, thanks for watching and until next time, this is ticker symbol U. My name is Alex, reminding you that the best investment you can make is in you.
Summary
- Nvidia reported record revenues of $96.2 billion, up 106% year-over-year, with a 66.2% operating margin.
- AI spending is accelerating, with Nvidia's revenue growth rates increasing over the past four quarters.
- The company now reports data center revenues separately for hyperscale and ACIE (AI clouds, industrial, and enterprise), with ACIE growing 138% year-over-year.
- CoreWeave, Nebius, and IREN are key players in the neo cloud market, with significant contracts from major companies like Microsoft and Meta.
- CoreWeave leads in scale and revenue but has high net debt and interest expenses.
- Nebius shows rapid growth and a strong balance sheet, with a unique model of customer prepayments covering buildout costs.
- IREN is transitioning from Bitcoin mining to AI, showing impressive revenue growth but facing risks related to its pivot.
- Investors should consider metrics like enterprise value and contracted power when evaluating these companies for potential investment.
Questions Answered
What key insights can be drawn from Nvidia's recent earnings call?
Nvidia's earnings call revealed record revenues and significant growth in AI spending, indicating a robust future for the company and the AI sector.
How do Nvidia's partnerships impact its market position?
Nvidia's partnerships with major AI buyers and cloud providers position it favorably in the AI market, reducing risk and enhancing growth potential.
What makes CoreWeave a significant player in the AI infrastructure market?
CoreWeave's impressive revenue growth and substantial backlog of contracts highlight its critical role in supporting AI workloads and its strong relationship with Nvidia.
What distinguishes Nebius in the AI infrastructure landscape?
Nebius stands out due to its rapid revenue growth and a business model that relies on customer prepayments, reducing financial risk.
How do CoreWeave, Nebius, and Iren compare in terms of financial health and market position?
A comparative analysis reveals that while CoreWeave leads in scale, Nebius shows stronger growth potential with less debt and a more favorable funding model.