The period after the dot-com bubble including the financial crisis of 2008 were difficult years for Microsoft. The stock returned a mere 38% compared to Amazon’s 657.9% in the same time frame and Apple’s 5150%.
Microsoft’s trajectory changed when Satya Nadella, formally of the Azure division, became CEO in 2014 after working his way up through the company over the course of 19 years to president of the cloud business. The stock is up 1,000% in the ten years since Nadella took the helm using his multi-decade cloud experience to steer a remarkable turnaround from a corporate reputation mired in fighting open-source communities and anti-trust issues. Since Nadella became CEO, the returns in Microsoft’s stock have exceeded Amazon and is tied with Apple, as of writing.
The competitor Nadella faced in building Azure is arguably the toughest competitor in technology – Amazon Web Services (AWS); not only for the vendor lock-in qualities of cloud IaaS as migrating a tech stack is quite costly in both time and money, but also because AWS had the first mover advantage of a four-year head start. In the tech industry, a lead this long is considered insurmountable.
Over the past ten years, Microsoft strategically exceled by targeting the Fortune 500 with 85% running on Azure today. Retaining the Fortune 500 in the migration to the cloud was accomplished through hybrid computing where Microsoft was first-to-market on serving a mix of on-premise, private and public clouds for their large enterprise customers. As the leader in on-premise systems, Microsoft was perfectly positioned to win with hybrid architectures. The company took this a step further and undercut other services on prices across its suite of software and platforms to win aggregate, long-term contracts.
This past month, for the first time, Microsoft has announced it will be re-organizing its reporting segments, which will afford investors a better apples-to-apples comparison between Azure and AWS. According to Wells Fargo, the new Azure reporting segment stands at an estimated $62 billion as of June 2024, compared to $105 billion for AWS.
The lead we see from Microsoft today on AI revenue streams is critical enough and predictive enough that it points toward Azure surpassing $200 billion by 2028, catalyzed by the OpenAI investment, Copilot’s rapid integration into nearly every Microsoft software product, having the ace of spades — which is an operating system used in 72% of the world’s laptops and desktops, and perhaps the simplest reason of all —- Microsoft excels at the enterprise.
New Azure Reporting Reveals 11-Points of AI Contribution
Microsoft offers investors unique insights as to the monetization opportunities for AI. Last year, in the FY2023 Q4 report ending in July, Azure officially inflected due to AI. Per the report: “Azure and other cloud services revenue grew 26% and 27% in constant currency, including roughly 1 point from AI services.” In four brief quarters, Microsoft is now reporting an 8% inflection from AI: “Azure growth included 8 points from AI services where demand remained higher than our available capacity.”
Later, it was stated in an updated FY25 investor presentation that Azure saw an 11 point contribution last quarter compared to the 8 points previously reported. The metric change is due to Microsoft removing Enterprise Mobility + Security (EMS) and Power BI (BPP) from Intelligent Cloud. It’s significant that Azure is seeing low double digits while AWS and Google Cloud are not reporting their exact contribution from AI, rather are remaining vague by saying “several billions” in AI revenue.
The reporting changes also update Azure’s growth rate to 33% for the fiscal year and an impressive 35% growth in constant currency. Prior to the metric changes, management guided for a slight deceleration in Azure growth in Q1’25, with growth of 28% to 29% in CC (vs 30% this quarter), yet they expect an acceleration in H2’25 as their capital investments increase AI capacity.
According to Wells Fargo, the new metrics suggest an annualized run rate for Azure of approximately $62 billion. Investors will get the official number in next quarter’s earnings report.
Management has stated the primary issue is being capacity constrained, which all things equal, is bullish for the medium-term as it implies demand exceeds supply for Azure AI and Azure’s consumption business. Per management in the most recent earnings call: “And in H2, we expect Azure growth to accelerate as our capital investments create an increase in available AI capacity to serve more of the growing demand.”
Azure AI is a platform for developing custom AI applications and solutions. Companies use Azure AI to integrate generative AI and multimodal language models into their applications for features such as search, image recognition, natural language processing, speech to text and other AI features using developer tools, such as APIs and SDKs. The platform also offers lifecycle management for data preparation and model development and training for machine learning, supporting popular frameworks PyTorch and Tensorflow. Azure OpenAI provides access to OpenAI’s GPT-4, GPT-3.5, Microsoft’s DALL-E models, and Meta’s Llama models for companies to build custom generative AI applications and AI assistants. Companies run models on their data to improve workflows through Azure AI Studio.
Azure AI customers totaled more than 60,000, implying customer growth rate of nearly 60% YoY and up over 13% vs Q2’24 with average customer spend continuing to grow. The number of Azure AI customers using data and analytics tools also grew nearly 50% YoY.
Where Azure stands apart is that its security segment is one of the largest in the world. In 2023, it was stated Microsoft’s security segment was at $20 billion with 860,000 customers. The number of customers has been updated to 1.2 million, and if we do some simple math, that would imply the security segment is at $28 billion today – far exceeding all best-of-breed cybersecurity companies combined.
Also tied to Microsoft’s strong presence in security, the Federal Government often gets overlooked in terms of its AI impact to Azure. In a blog post, the company CTO Bill Chappell wrote: “[…] generative AI capabilities through Microsoft Azure OpenAI Service, can help government agencies improve efficiency, enhance productivity, and unlock new insights from their data. Many agencies require a higher level of security given the sensitivity of government data. Microsoft Azure Government provides the stringent security and compliance standards they need to meet government requirements for sensitive data.”
Key metrics for Microsoft have been on fire lately. Bookings increased 17% YoY and 19% on a constant currency basis. This was significantly above expectations and driven by growth in the number of $10M+ and $100M+ contracts for Azure and Microsoft 365. This compares to 29% growth (31% on CC basis) in Bookings last quarter and compares to a -2% decrease (-1% on CC basis) in Bookings in the year ago quarter. Commercial RPO grew by 20% YoY to $269 billion. This compares to 20% growth last quarter and 19% YoY growth in the year ago quarter.
Source: I/O Fund Stock Research
If Azure were to continue its growth rate today on the assumption that any acceleration from AI offsets a deceleration on traditional cloud revenue (due to repatriation from moving cloud workloads to on-prem, for example), then Azure would reach revenue of $178.3 billion by 2028. That’s the bare minimum base case.
If we assume that AI contributes an additional 10 points for the next two years, and then tapers off to 8 points of AI contribution, and finally 4 points of AI contribution due to a higher revenue base, while also offsetting up to a 6-point decline in traditional cloud workloads, then Azure will reach $206.7 billion by FY2028 (ending in June of 2027).
Source: I/O Fund Stock Research
There are some analysts forecasting $41.6 billion for AI revenue for AWS by 2027. It’s reasonable to assume Microsoft’s AI revenue will be higher as it’s the only company reporting details on AI revenue across the Big 3 and at a double-digit percentage of 11% nonetheless (or about $7 billion in AI revenue) compared to vague comments of “several billions” of AI revenue from Azure’s competitors, and likely to be the $3 to $4 billion range. Therefore, assuming Microsoft has $55 billion in revenue by 2027 compared to AWS’ $42 billion is a reasonable assumption.
Copilot’s Rapid Integration
While OpenAI’s Chat-GPT and Google’s Bard /Gemini have gotten all of the attention, Microsoft has been quietly building an AI software empire with Copilot. Copilot features are designed to boost productivity and are integrated across consumer and enterprise software, including Windows, Edge, Office, Bing, Teams, Loop, Dynamics and Viva.
Copilot utilizes large language models that require Azure consumption. To prepare for this moment, Microsoft invested $1 billion in Open AI in 2019. Over the last five years, Microsoft has increased its investment to $13 billion. Open AI’s Chat-GPT are some of the large language models that power Microsoft’s Copilot. OpenAI leads the market on LLMs and every time Chat-GPT is integrated into a product and used across OpenAI’s user network, money funnels to Azure as Microsoft is the exclusive cloud provider in exchange for allowing OpenAI to access Azure’s infrastructure at a reduced cost. Around the time that Chat-GPT was noticed by Wall Street, Microsoft’s management team said the following about its impact on Azure:
“Second, even Azure OpenAI API customers are all new, and the workload conversations, whether it’s B2C conversations in financial services or drug discovery on another side, these are all new workloads that we really were not in the game in the past, whereas we now are.”
Developers pay between $10 to $19 per month for GitHub Copilot. According to the most recent earnings report, Copilot accounted for over 40% of GitHub’s revenue growth this year and is already a larger business than all of GitHub when Microsoft acquired it at $2 billion annual recurring revenue (ARR). GitHub Copilot has been adopted by over 77,000 companies, up 180% YoY.
Copilot Studio, a low-code tool for creating and maintaining copilots, saw a 70% QoQ increase in organizations using it to 50,000.
Copilot for Sales and Service is priced at $50 per user each, Copilot for finance is $30 per user, and for Dynamics data platform use cases, it’s priced as high as $1,397 per tenant. The highest cost is Copilot for Security, priced at $4 per hour, with an estimated monthly cost of $2920.
Although some of the upside from Copilot will be reported in other revenue segments outside of Azure, there are some tie-ins regardless of where Copilot is running. For example, enterprises need data to reside where Copilot can access it, which implies higher Azure revenue. On a more granular level, those who subscribe to Office 365 are often locked-in to Azure Active Directory (AD).
Today, Microsoft has more than 400 million 365 Commercial customer seats and 78.4 million 365 Consumer subscribers, giving a nearly ~480 million customer base to target for AI services. Assuming AI PCs help to spark a strong growth trajectory for Copilot, just a 10% adoption rate across both Commercial and Consumer by the end of the fiscal year would surpass $17 billion annual run rate. It’s likely the 10-year adoption rate will be well above 50% with an addressable market of up to 90% as more AI assistant productivity hacks are developed. This means Office 365 would land somewhere between $86 billion and $156 billion, equal to the current size of Azure or up to double the size of Azure on this revenue stream alone.
While Wall Street is worried about how much AI is costing, the I/O Fund is busy calculating how big the AI opportunity can get in the next few years and how investors can participate. Join our next webinar on Thursday September 12th where the Portfolio Manager will discuss potential entries for Microsoft and other AI-related stocks.
Copilot for Microsoft 365 is priced at $30 a month. The productivity tool combines large language models (LLMs) with the data in Microsoft Graph and Microsoft 365 apps. The use cases of Copilot in Word include giving users the first draft while saving the time on sourcing, writing, and editing the content. Similarly, Copilot in PowerPoint will help to create presentations based on previous content. Copilot in Excel can analyze trends from the data, create charts, and assist in making informative decisions.
In the second full quarter of availability, the number of people using Copilot for Microsoft 365 nearly doubled QoQ. Copilot customers increased 60% QoQ and the number of customers with over 10,000 seats more than doubled QoQ.
Power Platform, a collection of low-code development tools, saw MAUs rise 40% YoY to 48 million. 480,000 organizations have also used the AI-powered capabilities in Power Platform, up 45% QoQ. As stated, Power Platform will no longer be reported in Microsoft’s Azure segment
Copilot on the Verge of Becoming Ubiquitous
Windows operating system is in 72.3% of desktop and laptops. When you consider that Windows has most recently launched on an Arm-based PC with Qualcomm, and will also launch next year with AMD and Nvidia, that market share is likely to grow rather than contract. Similar to the penetration rate of Office 365, Windows dominates PCs regardless if the OEM is Dell, HP, Lenovo, Acer, Asus, LG, Samsung or Microsoft Surface.
Copilot+ for Windows is a sidebar that helps Windows users change settings, find files or summarize text across a desktop. Recall is a feature that helps a Windows user find documents, emails and web pages when a user simply states how they recall the file or digital asset. On some Surface laptops, there is a Copilot key, a digital pen enhanced with AI, sound and voice features enhanced with AI, and enhanced AI cameras.
Both Arm-based and x86 AI PCs are ramping this year with a CPU + GPU + NPU combo that will, in turn, proliferate Copilot+ for Windows, the AI assistant that requires a minimum of 40 TOPS (trillions operations per second). The neural processing units (NPUs) are powerful enough to deliver the official kickoff of AI edge computing as Copilot+ runs AI tasks locally on the AI PC and integrates them into various applications. This is an important moment for AI as prior to NPUs exceeding 40 TOPS, workloads were primarily sent to the cloud. By running AI assistants locally on the computer, suggestions will be faster and more accurate.
Qualcomm’s Snapdragon X Elite and Plus processors were released this last summer and offer the first GPU, CPU and NPUs that exceed 45 TOPS for AI tasks for Microsoft’s Copilot+ with a long battery life of over 12 hours. This week, Intel released its second-generation Core Ultra chips capable of reaching 48 TOPS that offers a long battery life of over 10 hours with the added benefit of running legacy x86 software without compatibility issues.
Canalys is projecting AI PC shipments to rise at a 44% CAGR from 2024 to 2028, from an estimated 48 million PCs this year, before doubling to more than 100 million in 2025 and rising to over 205 million by 2028. Cumulative shipments of AI PCs are projected to surpass 600 million over the next four years. I’ve covered additional information on the growth of the AI PC market here.
Within this rapid growth, commercial adoption is forecast to be higher, at approximately 60% by 2028 versus 40% for consumer. This is due to the productivity gains that AI PCs can enable via powerful on-device AI as well as benefits to software developers and related roles. For example, Dell’s XPS and Latitude 7455, equipped with the Snapdragon X Elite can support 13 billion-plus parameter models which means customers can run popular models like Llama 3 directly on their PCs. The fact that commercial adoption will be higher than consumer adoption is a boon for Microsoft Copilot and its suite of enterprise AI-enabled applications and platforms.
Microsoft’s Capex Spending Highest Among Cloud IaaS Providers
Microsoft is unabashedly spending tens of billions on AI infrastructure. In the last earnings report, the company announced strong QoQ increase to its capex for AI infrastructure. Capex was $14 billion last quarter, when it grew 22% sequentially. Microsoft’s capex increased 36% sequentially and 78% YoY to $19 billion in Q4.
Full year 2024 capex was up 75% YoY to $55.7 billion, yet this quarter’s run rate suggests we could see up to $80 billion in capex in FY2025. Compare this to cloud IaaS leader AWS which reported H1 capex of $30.5 billion for a run rate in capex of just over $60 billion. Notably, management is guiding for a further YoY increase in capex in FY’25. We’ve covered the importance of Big Tech’s capex for our AI stocks in an analysis here and also in a previous webinar.
Big Tech management teams have been getting an earful from Wall Street on when investors can expect to see a return. Private investors are busy calculating what level of revenue these companies must generate, estimating the return will need to be as high as $600 billion to justify the revenue Nvidia has reported in its data center segment.
Therein lies the disconnect, which is that Microsoft’s CFO, Amy Hood, states they are capacity constrained – implying the opposite problem, that the capex they’ve allocated is not nearly enough to serve Azure AI demand. Per the last earnings call: “We are – and we’ve talked about now for quite a few quarters, we are constrained on AI capacity. And because of that, actually, we’ve, to your point, have signed up with third parties to help us as we are behind with some leases on AI capacity. We’ve done that with partners who are happy to help us extend the Azure platform, to be able to serve this Azure AI demand. And you do see us investing quite a bit as we’ve talked about in builds so that we can get back in a more balanced place.”
Microsoft’s management team also pointed out about 40% of capex is spent on land, which is a long-term asset, and the rest is tied to a demand signal for inference. The CFO stated: “even in the capital spend, there is land and there is data center build, but 60-plus percent is the kit, that only will be bought for inferencing and everything else if there is demand signal.”
Therefore, investors have an important decision to make. On one hand, investors could listen to the bearish undertones that high capex spending will not be returned to investors over time, or on the other hand, view high capex spending as a bullish signal of the overflow in demand that will sustain for many years to come, with AI consumption well exceeding the capex being spent to build the infrastructure.
Conclusion:
In 2022, I wrote an 8,300 word analysis entitled Special Report: The New Kings of Tech for our research members that tied together key points on how to position our readers for AI’s big moment – well in advance of Nvidia’s stock surge. Part of this analysis was to emphasize that our readers should shift their mindset from consumer-facing stocks to enterprise-facing stocks. This is not easy to do given the FAANGs are primarily consumer stocks, and it was consumer that drove historic gains for the market over the past decade.
Here is what I wrote at the time:
“The adage is that history rhymes but it does not repeat. I believe a large addressable market is certainly required to produce the new wave of FAANGs – however, rather than consumer driving the gains, I believe it will be enterprises. Below, I discuss the enterprise-level market that will be four times larger than mobile and two stocks that will directly participate. Imagine participating in 4X the FAANGs by 2030. That’s what I believe will happen due to one key trend and I discuss exactly why this will be achieved below.” -I/O Fund’s Special Report: The New Kings of Tech, June 5th 2022
Nowhere will the AI enterprise advantage be more evident than with Microsoft’s steady ascent over the next ten years, which I believe will end with Microsoft firmly on top in nearly every category the company competes in. A few years ago, I predicted Nvidia would Surpass Apple by 2026. At the time, Nvidia had a $550B market cap and the mere thought was inconceivable . To that point, I purposely did not say Nvidia would surpass Microsoft — as once the AI opportunity fully plays out —- this company will be a tough one to catch.
Microsoft has been a heavy spender on AI infrastructure, with positive read throughs for AI semiconductors we own and a handful that we are tracking to buy soon. Learn more about the I/O Fund’s holdings and consistent deep dive research on AI stocks, crypto and more here
The I/O Fund does not own Microsoft at time of writing but is carefully planning our next entry. Real-time trade alerts are sent to research members and entries are discussed weekly in our webinars.
If you would like notifications when my new articles are published, please hit the button below to “Follow” me.