Recently leaked court documents during the Microsoft Activision hearing require us to revisit our cloud forecasts and market share data. The poorly redacted docs, which have since been removed from public viewing, suggest that Microsoft’s Azure revenue is at least 25% lower than our previous estimates. As a result, we’ve cut and revised our Azure revenue figures which in turn increases AWS’ big 4 hyperscale cloud market share. Our new estimates show that AWS maintains a greater than 50% share of revenue through 2023. While the change also helps Google Cloud, its market share is only modestly affected.
In this Breaking Analysis we update our hyperscaler cloud revenue estimates and market share data. We’ll also explain how the ETR data on cloud should be interpreted in this context and look forward to potential catalysts for cloud growth, including acceleration in Q4 attributable to generative AI.
Don’t Use a Sharpie to Redact Documents
The story was well covered in the press. The media picked up on Google analysis in late 2022 on a similar matter but the Activision hearing let the cat out of the bag…oops!
Last month as part of the FTC’s attempt to block the Microsoft Activision deal, Microsoft submitted redacted documents to the court that pegged its Azure revenue for its fiscal 2022 at $34B. That’s the four periods from June 2021 to June 2022. More than $10B lower than most analyst estimates, including ours.
How are we to interpret this data? On the one hand, it helps Microsoft to show data that it has significant competition in the cloud and doesn’t have a monopoly. Further, Microsoft can basically put revenue in whichever bucket it wants. Moreover…no one can seem to get their hands on the poorly redacted documents to see the exact context. For example, here’s some detail we’d like to see. Microsoft claims to have a $20B cybersecurity business. Is any of that revenue counted in Azure? If so how much? If not why not?
Regardless of the many questions, this revelation can’t be ignored. It’s highly unlikely that Microsoft would produce data in court documents that would not be consistent with how it reports its business internally. So we have to adjust our cloud figures.
The Opacity of Cloud IaaS and PaaS Reporting
Before we get into the revised data, let’s remind you that only AWS reports its IaaS and PaaS data in a reasonably clean manner. Microsoft reports Azure growth rates and Alphabet gives hints about its IaaS and PaaS businesses. But neither gives specific revenue figures. Below we show some examples. 
In this past earnings print, Microsoft gave more information than it ever has regarding Azure. The company cited that Azure’s annual revenue, for the first time, surpassed 50% of the Microsoft Cloud total, a figure it always reports. Microsoft consistently gives specific figures on Azure growth rates each quarter. Taking those into consideration as gospel and pinning FY 2022 at $34B (the leaked figure), we can get close to 50% but not quite. So we’re left asking ourselves, which figures are correct? The quarterly growth rates or the most recent disclosure? And how reliable are the leaked numbers.
We’ve always been dubious of the $34B leaked figure and felt the number was larger. But given the lack of transparency we’ll err on the conservative side and have decided to reduce Azure’s revenue in our model.
Revised Hyperscaler IaaS & PaaS Figures Shifts Market Shares
So what does this mean for our previous forecasts? Let’s take a closer look at the data.
Above, on the left hand side of this graphic, we show our previous estimates for the big 4 hyperscalers. The top chart is worldwide IaaS and PaaS revenue and the bottom is market share data prior to the leaked documents.
On the right hand side we show the same data for our revised estimates. Let’s review the meaningful changes:
- The Azure data has dropped considerably by more than 25%. Normalizing Microsoft’s fiscal years into calendar years, Azure revenue goes from $55B to $40B in 2022.
- As such, Azure’s market share in that period drops from around 35% to 28%.
- Meanwhile the impact on AWS is meaningful. While you see on the lower left we had AWS dropping to 50% in 2022 and below 50% in 2023, the revised estimates indicate that AWS had a 55% share in 2022.
- We expect that share to hold up in the low 50’s – 53.2% in 2023.
- While the change helps GCP look better – it’s not nearly as significant due to its smaller base.
Lastly, the overall market size in our model has been reduced by approximately 10%. We’ve revised our Big 4 2022 revenues down to around $145B growing 18% to $171B in 2023.
Why Such a Large Gap in Forecasts?
You may be wondering how could the estimates published in Breaking Analysis and other Wall Street numbers be so far off? The simple explanation is we’ve relied to heavily on customer data. When asked about cloud usage, customers report a large portion of their SaaS spending as cloud. Our ETR partners survey quarterly and ask more than 1,700 IT decision makers about their spending on cloud and other sectors. The chart below shows this dynamic in more detail.
The data above shows Net Score or spending momentum on the vertical axis and the penetration within the ETR data set on the horizontal axis. The table insert at the bottom right shows the KPIs for each company and informs the position of the plot.
Microsoft is shown in the upper right. It consistently shows the strongest performance in the ETR data set. As we’ve explained before, we’ve always suspected that much of that performance is due to the previous explanation – i.e. customers including SaaS spend in the cloud category. While we’ve consistently tried to correct for this anomaly in our forecasts, if the court leaks are to be believed, we have to concede we didn’t adjust the figures enough to the downside.
Note: We’re still not convinced the court documents are an accurate reflection of Microsoft’s Azure revenue. If you pin Azure’s FY ’22 numbers at $34B (the leaked figures) and model against Microsoft’s stated Azure growth rates, you can’t quite get to the implied $55B in FY ’23 that Microsoft just disclosed. In fact, if you ignore the leaked figure and model on the 50% disclosure, it suggests that our original forecasts were within 10% of Microsoft’s recently disclosed figures.
But given the lack of transparency, for now, we’ll go with the lower Azure figures.
Blurring Lines of Cloud: IaaS, PaaS, SaaS & On-Prem
As further evidence supporting our supposition, Salesforce shows up in the above graphic. In addition, as we’ve shared in previous research, Snowflake bundles AWS infrastructure costs in its service fees. It’s why we suspect that customers cite database costs as the most prone to surprises on the cloud bill. The strength of a time series survey like ETR’s is its consistency over time, which gives analysts the ability to identify and interpret changes. ETR is constantly evolving its taxonomy and granularity and has some plans to address such anomalies in the future.
As well on the above graphic there are a number of other firms and services that show up in the ETR data set. Including the functions offerings from all three big cloud players. Cloudflare has also begun to show up in the cloud computing sector of the taxonomy and has strong momentum as it builds out a new type of secure network. VMware – both VMware Cloud on AWS and VMware Cloud Foundation (VCF) are prominent as is OpenStack, which is popular for so-called sovereign clouds. Oracle and IBM have always had a presence in the data.
New names are also showing up in the data, notably HPE with GreenLake and Dell with APEX. While their spending velocity is not high, both companies have a reasonable N in the data with Dell somewhat higher in presence and HPE higher with spending velocity.
Lastly we show the Alibaba data as it’s another data point. It’s questionable how much the ETR data reflects Alibaba’s actual market performance but directionally it’s come down dramatically in the data set which aligns with Alibaba’s troubles in China.
Key Milestones to Watch in Cloud
Let’s talk about some of the things we’re watching in the space and some potential catalysts that could move the market.
Amazon reports earnings on August 3rd. We expect that cloud deceleration will moderate and AWS growth might actually accelerate a bit sequentially.
In addition, we’re working with the following planning assumptions:
- We see Gen AI as a tailwind for the cloud, particularly the big three cloud players. The big three clouds all have significant presence in the AI sector. AWS and Google have always had a technology lead and now Microsoft has “cut the line” as we’ve talked about with OpenAI. It certainly has commanded the Mindshare.
- We think that the deceleration in cloud growth rates reverses this year. We’re seeing signs of acceleration at Google. Microsoft Azure’s growth is decelerating but flattening. We’re assuming that in Q4 ,AI monetization starts to become more meaningful and that will accelerate cloud growth.
- Cloud growth continues to dramatically outpace the overall IT spending rates, which are hovering around 3%. Cloud is significantly higher than that in the high teens. But as we’ve reported, there’s a more balanced market vis-a-vis on-prem, versus the cloud.
We’re not Joining the Repatriates Yet, but the Market is Balancing
We’ve reported in our research that rebalancing and cloud optimization. We’re not repatriates just yet, but we have talked about been watching the Charles Fitzgerald (Fitzy’s) repatriation index. It measures the revenues of Digital Realty plus Equinix and divides by AWS revenue. The index bottomed last quarter and bounced up slightly. We’ll see how it does this quarter. We would suspect it’s not going to show a dramatic reversal but data suggests we’re hitting a sort of equilibrium.
On this last point we want to call your attention to a piece of research that was recently published by Dion Hinchcliffe of Constellation Research. He interviewed more than 20 CIOs. As our research has indicated, the market is becoming more balanced between cloud and on-prem. Cloud is still growing significantly faster but work that isn’t in the cloud will be tougher to move. Why? Because: 1) The cloud operating model is coming to a data center near you – i.e. the on-prem guys are leaning into as a service like APEX and GreenLake; 2) Their offerings are more robust than Outposts and better suited for the on-prem installed base; and 3) The work that is on-prem has characteristics that favor keeping the data where it is.
That’s what Dion’s research articulates very nicely and we show below.
We won’t in depth here but Dion did a very nice job of laying out the decision points for where work should be placed with some industry considerations and examples of notable use cases. And he provides a framework for understanding the salient characteristics of work that is best suited on-prem vs in the public cloud. We would encourage you to take a look at the research.
Gen AI as a Catalyst for Q4 Growth
As we said earlier, despite this state of balance, we believe generative AI will be a tailwind for the hyperscalers. All companies leaning in will benefit but we think the hyperscalers are best positioned. AWS with Sagemarker and Bedrock; and its custom silicon and other tooling; Google with its AI prowess and Microsoft which jumped to the mindshare lead with OpenAI. These platforms are all prominent in AI today. They don’t have to “AI wash” to be in the game because they’ve been in it for a while. As such, we expect generative AI will benefit cloud consumption for new workloads.
Further, we believe that the big 3 clouds in the U.S. will have dominant positions in LLMs. And others like OpenAI and Anthropic, etc. will play. But there will only be a handful of giant LLMs in our view. There will be many more smaller models that use proprietary data in various industries and use cases.
In addition, there’s a business case for AI coming onon-prem work so we’ll be watching that closely. ETR data shows that 96% of companies in a recent survey either have or are formulating an AI strategy. Sixty-four percent (64%) say it will increase their infrastructure needs. And 30% say it will accelerate their cloud spend. So this is something we’re watching. Does that gap suggest that we’ll see a resurgence on on-prem / hybrid spend around AI.
The key will be the degree to which the traditional players can integrate generative AI and establish credibility in the market. As well, how fast they can actually build these AI capabilities into their as-a-service offerings; and build partnerships with teeth.
“AI or Die” is the new slogan for all.
Keep in Touch
Many thanks to Alex Myerson and Ken Shifman on production, podcasts and media workflows for Breaking Analysis. Special thanks to Kristen Martin and Cheryl Knight who help us keep our community informed and get the word out. And to Rob Hof, our EiC at SiliconANGLE.
Remember we publish each week on Wikibon and SiliconANGLE. These episodes are all available as podcasts wherever you listen.
Email david.vellante@siliconangle.com | DM @dvellante on Twitter | Comment on our LinkedIn posts.
Also, check out this ETR Tutorial we created, which explains the spending methodology in more detail.
Watch the full video analysis:
Image: JohnKwan
Note: ETR is a separate company from Wikibon and SiliconANGLE. If you would like to cite or republish any of the company’s data, or inquire about its services, please contact ETR at legal@etr.ai.
All statements made regarding companies or securities are strictly beliefs, points of view and opinions held by SiliconANGLE Media, Enterprise Technology Research, other guests on theCUBE and guest writers. Such statements are not recommendations by these individuals to buy, sell or hold any security. The content presented does not constitute investment advice and should not be used as the basis for any investment decision. You and only you are responsible for your investment decisions.
Disclosure: Many of the companies cited in Breaking Analysis are sponsors of theCUBE and/or clients of Wikibon. None of these firms or other companies have any editorial control over or advanced viewing of what’s published in Breaking Analysis.
