Microsoft: AI Copilot Factory
Microsoft: AI Copilot Factory
Q1 2024 ER Update
This newsletter is sponsored by me! More than 20 students have already enrolled in my 2 Hour Deep Diver course, in which I teach you how to analyze companies in depth on your own, like I do.
By completing the course, you will emerge as a self-sufficient analyst with a distinct ability to do more research in less time, know when to buy or sell a stock and more easily spot undervalued companies, among other valuable skills that will help you compound wealth over your life, slowly.
You will receive personalized attention from me, for free, via the course’s Slack channel.
For just $150 and in under two hours of (highly compressed) learning material.
Edited by Brian Birnbaum.
In Section 1.0 I explain how Microsoft is expanding margins by being smart about its architecture.
In Section 2.0 I explain how Microsoft is set up to increase its earning power by nurturing and connecting ecosystems.
In Section 3.0, I discuss how the automation space is heating up, with players like Nvidia, Palantir and Microsoft in the same battlefield.
In Section 4.0 I analyze Microsoft’s financials and in Section 5.0 I conclude the write up.
1.0 Margins Driven by Architecture
Seen as a unified server, Microsoft has made excellent progress with creating the infrastructure required for success in the era of AI. This is evidenced by the expanding margins seen this quarter.
Traditional analysts like to see Microsoft as a collection of segments, but this impedes a truly productive understanding of the company.
At this stage, Microsoft is essentially an AI copilot factory, as Satya explained in the conference call:
We're using this AI inflection point to redefine our role in business applications. We are becoming the Copilot-led business process transformation layer on top of existing CRM systems like Salesforce.
It is a unified server that dishes out business applications to billions of people worldwide. As folks use these apps, they generate data, which can then be used to train AIs that automate work.
In turn, Microsoft enables organizations to rent the computing infrastructure that the company uses to operate its business applications in the first place (Intelligent Cloud segment).
Microsoft uses its edge at the operating system level (More Personal Computing segment) to distribute business apps worldwide (Productivity and Business Processes segment), which then drive data generation.
Below you can see how Microsoft’s business segments emerge from the OS layer; you’ll notice that revenue within the “More Personal Computing” segment is shrinking in percentage terms as time progresses.
Once you work with a copilot for the first time, there’s no going back. It is a fundamentally improved way of working, akin to having electricity at your disposal or not.
While of course business applications like Microsoft Word can be intrinsically improved over time, the “killer feature” is having an AI that does the work for you.
Going forward, for Microsoft to meaningfully increase its earning power, it must create an infrastructure that enables:
The continuous deployment of new copilots and improvement of existing ones.
One model to run many copilots, in any Microsoft app, to maximize the leverage per AI model trained.
Per the results seen this quarter, this is exactly what Microsoft has been working on of late.
Microsoft’s gross margin came in at 71.16% in Q1 FY2024, up from 69.84% last quarter–a high since 2014.
In turn, operating margin came in at 47.59%, up from 41.08% last quarter [1].
According to management, increases in gross margin are due primarily to ‘improvements’ in the cloud and Office 365 businesses. Satya clarifies these improvements during the Q&A:
But the thing is, we have scale leverage of one large model that was trained and one large model that's being used for inference across all our first-party SaaS apps, as well as our API in our Azure AI service…
The lesson learned from the cloud side is–we're not running a conglomerate of different businesses, it's all one tech stack up and down Microsoft's portfolio, and that, I think, is going to be very important because that discipline, given what it will look like for this AI transition, any business that's not disciplined about their capital spend accruing across all their businesses could run into trouble.
Over time, this architecture will enable Microsoft to maximize the number of users engaged with copilots daily, while minimizing computing expenses. This should ultimately equate to a higher earning power.
The same architectural configuration that enables Microsoft to do this is also very appealing for Intelligent Cloud customers because they all need to do the same with their businesses.
CFO Amy Hood ratifies this:
And the reason, Mark, that's important is that it means, even beyond the point Satya made is that, when it comes to our ability to leverage the infrastructure that we're building out, we don't really have a preference in terms of how people are utilizing that infrastructure, whether it's through all the means that Satya mentioned.
It gives us a good opportunity to see quick conversion into revenue.
2.0 Value Creation by Nurturing Ecosystems
With its proven ability of nurturing and proliferating ecosystems, Microsoft is poised to produce datasets that will redefine how we work and truly leverage the above architectural configuration.
As I explain in my Microsoft deep dive, the company’s ecosystems exhibit fractality: the company has a number of large ecosystems, from which other smaller ecosystems emerge.
Microsoft’s culture and organization nurture these ecosystems without losing focus.
Doing so enables Microsoft to create novel datasets that allow the company to create competitive AI copilots.
For example, this is the first time since I started analyzing Microsoft that I’ve seen them taking meaningful steps to merge their Office 365 and LinkedIn ecosystems. The former provides tools for work and the latter a network that connects participants of the labor market.
The combination of the two ecosystems will yield a comprehensive environment for people to do their best work, which itself will generate vast amounts of data.
This data can then be used to train AI copilots for a broader range of work-related tasks than if the two ecosystems were disconnected.
For instance, Microsoft is now in a position to create a copilot that helps people navigate the labor market over the course of their professional lives. Microsoft knows what work you do and who you are connected to. It’s a natural step forward.
In February 2021, Microsoft launched an app called Microsoft Viva, which helps organizations create a more engaging and productive workplace.
Incidentally, LinkedIn now has 985M members, up from 900M in Q2 FY2023.
Viva is now ingesting data from both Office 365 and Linkedin:
With skills in Viva, we're bringing together information from Microsoft 365 and LinkedIn to help employers understand workforce gaps, and suggest personalized learning content to address it, all in the flow of work.
-Satya Nadella, Microsoft CEO during the Q1 FY2024 conference call.
Not all copilots that Microsoft creates will work, but so long as Microsoft maintains the dominance at the OS and business applications levels, each additional iteration increases their chances for success.
Microsoft will likely find a few copilots that will drive most of the returns.
3.0 Crowding the Automation Space
As Microsoft moves towards helping customers aggregate their data real estate, the automation space is getting increasingly harder to analyze. But there is one powerful mental model that I know I can rely on.
At this point, Microsoft, Palantir, Nvidia and UiPath are all pointing in the same direction: towards enabling automation for organizations. This is quickly becoming a highly crowded space, with competitive boundaries looking very blurry.
Microsoft has a clear competitive advantage over the other companies outlined above when it comes to renting out compute power.
As it makes strides toward helping customers aggregate their data real estate, its distribution prowess promises to be a great asset.
At the end of this quarter, 21,000 organizations are using Azure Arc to orchestrate migrations to the cloud (up 140% YoY). 73% of the Fortune 1000 is using Microsoft’s Intelligent Data Platform, which includes Microsoft Fabric–far more organizations than Palantir (453) and UiPath (10.6K) serve, for instance.
Microsoft Fabric is essentially something like Palantir’s Foundry.
In turn, Power Platform enables anyone to create apps to analyze data, using natural language. This is, at a high level, similar to Palantir’s AIP.
Simultaneously, I recently learned that Palantir is working on a feature that will enable organizations to automatically mine processes that can be automated, much like UiPath does.
And of course, Nvidia is helping customers create custom AI models that automates work across their respective organizations too.
As the space gets more crowded, there are two things that I can bet on today: not only distribution prowess, but also culture.
I have seen Microsoft do well in certain spaces, but not dominate. Take cybersecurity, for example. In this particular battle with Crowdstrike, despite Microsoft’s clear advantage on the distribution side, the latter has not been able to take a number one spot in the market.
This is because Crowdstrike has a series of architectural advantages that have enabled the company to get ahead. These advantages in turn stem from an exclusive focus on XDR that’s been hard for Microsoft to beat.
Over time, the exclusive focus has translated into a culture, from which Crowdstrike’s competitive advantages ultimately emerge.
Termed as theInnovation Stack by Block’s co founder Jim McKelvey, as Microsoft and others make their way towards automation, I continue to look for this dynamic.
4.0 Financials
AI is driving increased cash flow for Microsoft, and the company is in great financial health.
Income Statement
Total revenue for the quarter came in at $56.5B, up 13% YoY.
Microsoft’s three business segments are growing very healthily, but it’s the cloud business that’s driving cash flow:
Productivity and Business Processes revenue grew 13% YoY to $18.6B, driven byy ‘health’ renewal rates and higher ARPU (average revenue per user) in Microsoft 365 and higher than expected LinkedIn revenue growth (up 8% YoY).
Intelligent cloud revenue grew 19% YoY to $24.3B.
More Personal Computing revenue decreased 1.7% YoY to $13.67B, tracking the evolution of the broader PC market.
Gross margins increased 1%, 2%, and 5%, respectively, the latter driven by a ‘shift to higher-margin businesses.’
Although margin expansion in the PC business is impressive, margin expansion in the cloud is far more accretive to the bottom line due to larger volume.
The delta in the cloud business is driven by an increased GPU capacity and a higher GPU utilization, which, according to management, came in far above expectations.
This coincides with Jensen’s remarks during Nvidia’s Q3 FY2024 earnings results conference call.
Cash Flow Statement
Microsoft’s cash from operations is re-accelerating, driven by the cloud business as previously explained.
Seen from an annual perspective, the evolution Microsoft’s cash from operations over the past few years reflects a stalling B2B deal pipeline.
But seen in a quarterly context, Microsoft’s story is more about AI driving cash flow growth again. Q1 FY2024 cash from operations came in at $30.6B, up 32% YoY.
With a CapEx of $11.2B for the quarter, including finance leases, free cash flow thus came in at $20.7B.
In the graph below you can appreciate how Microsoft has meaningfully stepped up its CapEx levels in the past two quarters. According to management, said increases are primarily directed at supporting increased cloud demand.
Balance Sheet
Cash and equivalents came in at an incredible $143.9B. Long term debt came in at $41.9B, giving the company an ample net cash position.
Notably, short term borrowings spiked from $0 to $25.8B QoQ, which is perhaps a source of concern. This was not mentioned in the call, which I find quite strange.
5.0 Conclusion
See a higher GPU utilization driving Microsoft’s cash flow is what I need to see to verify whether the AI hype was real. It seems as Jensen Huang points out in the latest Nvidia earnings report, AI copilots are really moving the needle.
Some months ago I had doubts about whether the current AI hype was going to move the needle, financially speaking. But now, I see that it has helped nudge Microsoft back on track for growing cash flows.
I look forward to analyzing the company over the coming few quarters, to see if the trend holds.
[1] Microsoft’s gross margin declined from 2010 (80%) to 2016 (62.86%) as it began to invest in the cloud business, which had significant infrastructure requirements.
⚡ If you enjoyed the post, please feel free to share with friends, drop a like and leave me a comment.
You can also reach me at:
Twitter: @alc2022
LinkedIn: antoniolinaresc