Cloud Wars Live with Bob Evans · 2026-08-11 · 6 min
Key moments - from our scoring
Substance score
44 / 100
Five dimensions, 20 points each
The four cloud hyperscalers have accumulated an unprecedented $2.3 trillion in contracted, signed revenue commitments - money already booked but not yet recognized on financial statements. Microsoft leads with $678 billion in RPO growing at 84%, followed by Oracle at $638 billion (363% growth), Google Cloud at $514 billion in backlog (339% growth), and AWS at $496 billion (154% growth). This fully committed business explains why capex spending from AWS and Google Cloud approaches or exceeds $200 billion annually, with all four companies increasingly turning to debt markets to fund data center expansion. While some Wall Street observers worry about temporary quarterly negative cash flows and perceived over-dependence on OpenAI and Anthropic partnerships, Evans argues this perspective misses the unprecedented scale and durability of the underlying enterprise demand. The hyperscalers are responding rationally by diversifying - AWS launched a $25 billion annualized chip business, and Google Cloud is selling TPUs to external customers - ensuring they're not solely dependent on any single AI vendor.
Microsoft, Oracle, Google Cloud, and AWS combined have $2.3 trillion in RPO (remaining performance obligations) or backlog - fully signed, contracted business not yet recognized as revenue.
The massive capex spending is justified by the $2.3 trillion in already-signed, contracted enterprise commitments in their backlogs, which requires building data center and AI infrastructure capacity to deliver those commitments.
No; temporary negative cash flow quarters reflect rational capital allocation decisions to fund infrastructure for already-committed revenue, not unsustainable spending, particularly given the unprecedented scale of their contracted backlogs.
While there is clear dependence on OpenAI and Anthropic, the hyperscalers have strong, independently growing cloud and AI businesses outside these partnerships, and are diversifying through chip manufacturing and TPU sales.
AWS launched a chip business with a $25 billion annualized run rate, and Google Cloud is selling TPUs to other companies to increase capacity and reduce dependency on single vendors.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode delivers solid data-driven substance with specific RPO figures and growth rates for the hyperscalers, but relies heavily on restating those numbers without deeply analyzing implications or business model mechanics. The insight about negative cash flow being contextually acceptable for market expansion is useful but not novel, and the discussion of AI dependency is acknowledged but not deeply explored.
$2.3 trillion in backlog or RPO...This is fully committed business...This is signed contracted business.
These are very well run companies...the perspective is being switched here. Traditionally the idea is bad, you don't want to have negative cash flow...The difference is there have never been a market with uh, a size, a total addressable market, anything like this growing at the rate this is.
The reframing of negative cash flow as contextually acceptable during hypergrowth is mildly counterintuitive but increasingly common commentary. The speaker mostly presents consensus cloud-industry narratives - capex races, AI chip development, RPO growth - without contrarian argument or first-principles thinking. The dismissal of OpenAI/Anthropic dependency concerns feels more like reassurance than original analysis.
Traditionally the idea is bad, you don't want to have negative cash flow. Okay, that's pretty basic. I think we got that.
These companies are so tied to OpenAI and Anthropic. If either OpenAI or Anthropic sneezes...There's clearly a dependence, a certain dependence on OpenAI and Anthropic. But...I also think it's being vastly overblown.
This is a solo host commentary, not a guest interview. No practitioner, operator, or expert guest is present to provide perspective or lived experience.
Speaker A: Hello my friends, welcome back to Cloud Wars Minute.
The episode is rich with named companies, specific financial metrics, and quantified data points: RPO figures ($678B Microsoft, $638B Oracle, $514B Google Cloud, $496B AWS), growth rates (84%, 363%, 3390%, 154%), capex targets ($200B+), and concrete business moves (AWS chip business at $25B ARR, Google selling TPUs). This specificity is the episode's strongest dimension, though interpretation of causal drivers remains vague.
Microsoft has an RPO, they call it $678 billion of contracted business. They're growing, that is growing at 84%. Oracle off a much smaller revenue base has this huge future business coming in. $638 billion in RPO growing at 363%.
AWS said, now we've got a, uh, chip business with an annualized run rate of $25 billion.
This is a monologue, not a conversation. There are no follow-up questions, pushback, or dynamic dialogue to evaluate. The speaker moves through talking points without probing deeper into assumptions, risks, or nuance. The format precludes conversational craft entirely.
Speaker A: Hello my friends, welcome back to Cloud Wars Minute. We've got another example here...
Computed from the transcript - who did the talking, and the words that came up most.
In today's Cloud Wars Minute, I analyze the extraordinary RPO growth at Microsoft, Oracle, Google Cloud, and AWS and what it signals about AI demand. Highlights 00:03 - We've got another example here where, in the greatest growth market the world has ever known, we are working with some big numbers that put the law of big numbers to the test here. So, if you look at the four hyperscalers, and I go in order of the size of their backlog or RPO , you've got Microsoft, Oracle, Google Cloud, and AWS. 00:29 - So this is fully committed business. It's fully contracted and not yet recognized as revenue. So this is what's coming down the road, to look into the pipeline, in the future, these companies have - this isn't some guesstimate of what they hope they'll get. This is signed, contracted business. So this is one of the factors, probably the key factor, behind why you see these CapEx numbers approaching or exceeding $200 billion. 01:09 - Now that has led to a couple of these companies entering into the debt markets to try to fund this data center expansion and all that enormous CapEx outlay that they've got to go through.
Transcribed and scored by The B2B Podcast Index.
Speaker A: Hello my friends, welcome back to Cloud Wars Minute. We've got another example here where in the greatest growth market the world has ever known, we are working with some big numbers that put the law of big numbers to the test here. So if uh, you look at the four hyperscales and I'll go and order the size of their backlog or RPO, you've got Microsoft, Oracle, Google Cloud and AWS now have amassed a total backlog or RPO of $2.3 trillion. So this is fully committed business, not it's fully contracted and not yet recognized as revenue. So this is what's coming down the road. Look into the pipeline in the future these companies have, but this isn't some guesstimate of what they hope they'll get. This is signed contracted business. So this is one of the factors, probably the key factor behind why you see these capex numbers approaching or exceeding $200 billion for Google Cloud and AWS. Microsoft approaching that Oracle on a smaller level but still increasing Cap X very, uh, very rapidly. Now that has led to a couple of these, uh, all of these companies entering into the debt markets to try to fund this data center expansion and all that enormous capex outlay that they've got to go through. And in turn a couple of these companies for a quarter to had negative cash flows. And that's got some people on Wall street just, you know they, they can't comprehend it, the world's coming to an end, what are we going to do? But I think uh, while that's not an ideal and I think these are very well run companies, very well financially managed companies, the numbers show that time after time after time that I think the perspective is being switched here. Right. Traditionally the idea is bad, you don't want to have negative cash flow. Okay, that's pretty basic. I think we got that. The difference is there have never been a market with uh, a size, a total addressable market, anything like this growing at the rate this is. Look at these latest numbers uh, for the four hyperscalers. So Microsoft has an RPO, they call it $678 billion of contracted business. They're growing, that is growing at 84%. Oracle off a much smaller revenue base has this huge future business coming in. $638 billion in RPO growing at 363%. Google Cloud had a huge jump this past, uh, Q2, $514 billion in backlog up 3, 390% and a resurgent AWS posted its biggest backlog number ever. $496 billion. And I believe that growth rate of 154% for its backlog is much higher than any they've reported over the last five or six quarters. Now, as a result of this, you see things like AWS said, now we've got a, uh, chip business with an annualized run rate of $25 billion. So they're moving more deeply in that way. Again, to help catch up with this demand, you've got Google Cloud saying, we're going to start to sell TPUs to other companies. They have, uh, figured they're going to be cranking that up, uh, to such a degree that there's a market to sell these elsewhere, all in pursuit of chasing these. Um, so there's the big question, I think for some folks, like, oh, my gosh, these companies are so tied to OpenAI and Anthropic. If either OpenAI or Anthropic sneezes, you know, the world is going to catch cold or pneumonia and die very quickly. There's clearly a dependence, a certain dependence on OpenAI and Anthropic. But these four companies also have very strong bases in cloud and AI outside of those two companies. So I think it's significant, but I also think it's being vastly overblown. I just want to say one more time, $2.3 trillion. Now, that's not like a TAM figure. Somebody's saying, oh, we think the market for new scooters is going to be $2.3 trillion. These are four companies just for, not the whole tech industry. And these are their signed, contracted, committed, uh, figures for what they've got in their backlog, or rpo. This just is completely unprecedented in an unprecedented time. Unprecedented approaches are often called for. So you see these moves into borrowing to find capex, you see these temporary dips into negative cash flow for a quarter. But both OpenAI and Anthropic have very, very big and rapidly growing enterprise businesses. And they're not these labs that are just out spending money but not bringing in any revenue. So, um, we'll keep an eye on this. I've got a lot more details in an, uh, article on cloudwars.com later this morning. I hope you'll check that out. Interesting way to, uh, you know, get rolling here in the, the next round of these, uh, extraordinary results here for what's going on in the greatest growth market the world has ever known, that. Thanks for being with us here at Cloud Wars Minute. I hope you have a fantastic day.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
Winning the Future: Google Cloud Beats AWS, Ties Microsoft Cl.Microsoft Hits 40 Million AI Agents: What It Means for Enterprise AIMicrosoft and Tech Leaders Push Back on Restrictions for Open-Weight AI ModelsPalantir Calls Out AI Labs as Q2 Soars 93%OpenAI's GPT-5.6 Becomes the Preferred Model in Microsoft 365 Copilot
