Microsoft: It’s A Marathon, Not A Sprint
Contributor since 2013 / 3115 Followers
Summary:
- Microsoft Corporation is betting big on Artificial Intelligence, or AI.
- It’s not certain that AI will deliver on its promise.
- Even if the AI phenomenon fails to take off, Microsoft will continue to deliver.
- Place your wager on MSFT, not AI.
- We’re increasing our MSFT Price Target to $485/share. Reiterate Buy Rating.
HJBC
Investment Conclusion
Microsoft Corporation (NASDAQ:MSFT) is an Artificial Intelligence (AI) front-runner. Driven by a large AI platform, with AI tools and services that developers like best, an enterprise focus, and OpenAI partnership, MSFT appears Golden. However, late at night, with the lights turned out, what is Satya Nadella, MSFT’s CEO thinking — AI is not in the bag yet, Amazon (AMZN), with its ample resources and superior skills, could build a better AI engine, and the OpenAI partnership could fizzle, taking access to exclusive, cutting-edge AI technology with it.
After all, as yet, AI has been good mostly for Nvidia (NVDA), with MSFT, AMZN, and Google (GOOG) holding the short end of the stick, buying NVDA’s ultra-expensive GPUs, with little to show for in AI derived revenues. In addition, Andy Jassy, AMZN’s CEO, in April, in his letter to shareholders, indicated that the company is going to use the primitives brush AMZN utilized to build its massive retail and cloud businesses, to develop the firm’s AI platform, and overtake the industry. Moreover, Open AI CEO, Sam Altman’s appetite for generating revenues is likely only to be further whetted by the firm’s success with commercializing its ChatGPT and AI plugins.
However, we’re taking a 34,000 feet view on MSFT. The company is far from dependent on the fortunes of AI. AI will always be around for MSFT, as the technology is integrated across the firm’s technology stack. If AI fails to deliver on the dreams most investors are dreaming, MSFT will still have Azure, Office, LinkedIn, GitHub, and gaming to pick up the slack.
To account for reasonable uptake of Copilots associated with Microsoft 365, Teams, and GitHub, as well as improved business dynamics, we’re increasing our MSFT Price Target to $485/share from the previous $420/share. To arrive at our new Price Target, we’ve updated inputs to the 10-year Discounted Cash Flow model, utilized to value MSFT. Specifically, our straight lined annual revenue growth rate goes to 16.5% from the prior 16%, normalized annual capital expenditure as percent of revenues escalates to 15% from the earlier 12%, and the weighted average cost of capital increases to 10% from the previous 9%, to account for greater risk to the story. The operating cash flows margin and perpetual growth rate remain unchanged at 46% and 3%.
Lay Of The Land
MSFT is all in – betting big on AI, having invested tens of billions of dollars in AI assets across the globe. Regarding AI hardware, along with the choicest AI chips from key semiconductor companies, MSFT has developed its proprietary Maia, AI accelerator, and Azure Cobalt CPU. Regarding Large Language Models (LLMs), along with offering a substantial selection from external firms, MSFT released several versions of its homegrown small language model Phi. Further, to support enterprise AI workloads, the company offers data management tools such as Graphs, Fabric, and the Power Platform. Moreover, to encourage developers to build on top of models it provides, MSFT offers a comprehensive suite of developer tools and services, including Azure AI Studio, Visual Studio Code, Azure SDK, Azure Developer CLI, Nvidia AI Enterprise, Nvidia AI Foundation Endpoints, and Nvidia GPU Accelerated Virtual Machines. Furthermore, the firm has integrated AI into most of its products, from Office to Windows, to Dynamics, LinkedIn, Teams, and GitHub. Undoubtedly, MSFT believes that AI likely represents the next leg of growth in technology, and is banking on it to deliver.
Developers play a critical role in the success of any technology. In that regard, it appears that MSFT has an edge. At GOOG’s I/O event earlier this year, programmers indicated that they preferred MSFT’s AI tool kit to GOOG’s, which they characterized as intimidating to approach.
An additional advantage is MSFT’s enterprise focus. From Office to Windows to Azure Cloud, a significant fraction of the company’s revenues are derived from sales to large corporations. With MSFT products already integrated into their technology ecosystem, partaking of the firm’s AI platform, appears to be the logical next step for these customers, from the perspective of saving time and money – who wants to transfer massive amounts of data to a new system, and deal with a steep learning curve. In addition, developers appreciate MSFT’S OpenAI connection, which offers them a steady supply of cutting edge new LLMs.
MSFT has in total invested more than $13 billion dollars (with a new funding round announced on Wednesday), in OpenAI, providing the company with supercomputing infrastructure, and Azure Cloud to power its research, products, and API services. In exchange, MSFT secured a 49% stake in OpenAI’s for profit arm (updated details on profits terms following the recent funding round are anticipated), until the firm invents Artificial General Intelligence (AGI), exclusive access to OpenAI’s LLMs, and an opportunity to integrate the company’s technology into MSFT’S products.
OpenAI’s technology has been embedded into MSFT’s Bing search engine, Edge browser, GitHub Copilot, Teams Copilot, and various Copilots associated with Microsoft 365. In addition, MSFT is the only cloud provider that has rights to offer OpenAI’s LLMs to customers. Moreover, considering OpenAI’s strategy to arrive at AGI in incremental steps, releasing models to mark the progress, provides MSFT with a steady supply of new AI models.
On the other hand, according to Sequoia Capital, AI ROI in 2023 has been dismal. The company estimates that for $50 billion spent in 2023 on NVDA’s GPUs, the industry returned $5 billion in AI revenues.
Regarding the competition, it is important to note that as AI is disseminated through the cloud, cloud computing market share is a key factor associated with AI revenues. AMZN’s AWS has a robust stable of AI models and tools. Its Sagemaker provides LLMs and tools for organizations to build their proprietary LLMs. In addition, AWS offers through its Bedrock service, LLMs customers can fine tune with their data. Q is a coding assistant that helps programmers write and debug code with relatively less effort and time. Industry commentary indicates that AMZN’s AI platform is highly suitable for AWS customers seeking to integrate AI into their businesses. Given that AWS represents 31% of the cloud computing market, AMZN appears to have a solid audience to upsell its AI platform. With 25% share of the cloud computing market, MSFT is not far behind.
Regarding GOOG, the firm offers Vertex AI, which provides many LLMs sourced from key technology companies as well as its proprietary Gemini and Gemma LLMs, as well as AI infrastructure and applications. With GOOG’s AI tools and services, customers can build their own AI models or customize offered LLMs with their data. Regarding AI applications, the company offers Gemini, a Generative AI (GenAI) conversational agent which generates text, images, and code based on natural language user prompts, and DuetAI as AI support for its office productivity platform. With just 11% market share in cloud computing, GOOG will need to provide customers a significantly superior AI experience to capture the market from the incumbents.
In additional competition, OpenAI, which previously derived most of its revenues from a premium version of its ChatGPT, has begun offering for fee AI services to businesses, ChatGPT plugins, and a web browsing application, positioning the firm as direct competition for MSFT.
It would only be fair to say that MSFT is for the most part in a state of waiting for the other shoe to drop at OpenAI, given the corporate shenanigans the company appears constantly to be embroiled in. Last Wednesday news broke that Mira Murati, OpenAI’s Chief Technology Officer (CTO) was stepping down from the organization, along with a couple of high-ranking executives associated with OpenAI’s research efforts. On Friday, it was reported that the company would shift its status to a for-profit from the previous non-profit, and Sam Altman would receive an equity stake.
Considering that OpenAI was founded by Elon Musk, among others, as a non-profit with the objective of averting GOOG’s AI dreams, which Elon Musk contended were a threat to human existence, the switch is remarkable. Over the years, a host of OpenAI founders including Elon Musk, John Schulman, Andrej Karpathy. Ilya Sutskever, Greg Brockman, and Jan Leike have left the company (only two of the original team of twelve remain, Sam Altman and an AI engineer), enabling the current juncture, with OpenAI probably eyeing an Initial Public Offering (IPO) in the near future.
Considering the scenario, it appears prescient that Satya Nadella spooked by the drama that unfolded at OpenAI in late 2023 (with the firing and rehiring of Sam Altman as CEO), brought in Mustafa Sulyeman (cofounder of GOOG’s Deepmind) and most of his team from Inflection AI, along with $650 million in intellectual property associated with the enterprise. Mustafa Sulyeman came in as CEO of newly founded Microsoft AI, assigned with the development of customer AI products. The maneuver was a possible attempt by Satya Nadella to hedge bets if the corporate happenings at OpenAI circumvented its AI potential, jeopardizing MSFT’s steady access to cutting-edge AI technology. MSFT’s additional AI investments include fund injections into Abu Dhabi based G42 and France’s Mistral AI.
The AI Conundrum
AI is the ability of computers to see, hear, and understand, spoken and written language as well as pictures. To harness that capacity, computer models are trained, sometimes on trillions of parameters and tokens of data, so they can respond to customer prompts based on patterns and statistical relationships that the user text or image invokes. The models are called LLMs and their output GenAI. Typical functions of GenAI include the generation of text, audio, video, and images based on natural language prompts, data analysis, data synthesis, translation, transcription, and making recommendations. ChatGPT, DALL.E, Sora, the Copilots, Siri, and Alexa are some examples of GenAI applications. For the most part, these chat with users in natural language; reproduce great works of art; rank order products; create emails, articles, and marketing material; write and debug code; analyze data; convert text to speech and vice versa; and translate text.
Of course, we were impressed when ChatGPT could converse with us. However, that was 2022. Based on a recent study, of 2,800 executives polled to determine AI ROI, 83% indicated that they used AI for report writing, summarization, and creating marketing copy; 62% for coding improvement; and 56% to facilitate audio conversations in call centers. $50 billion in NVDA’s chips for automating emails, translating and recording meeting minutes, and writing code, for the most part, appears excessive. Therefore, Sequoia Capital is not wrong in asking, what is going on? Even if the cloud hyperscalers are stacking compute for potential AI business, where are the discussions on serious future use cases for AI that will reflect in millions of dollars in cost savings for enterprises.
On a webpage dedicated to AI, MSFT indicates that significant effort will be required for organizations to integrate AI into their workflows. The company suggests firms stick with it, understand MSFT’s portfolio of AI tools and services through immersive experiences, identify use cases where AI would be useful to solve an issue, and then develop the requisite AI solution, deploying the right models and infrastructure.
Andy Jassy, in his recent letter to shareholders, draws parallels between the launch trajectories of AMZN’s retail and cloud computing segments and that of their AI business. He proposes that developing primitives, basically building blocks based on first principles, will be the way forward for AMZN’s to expand their AI business. Andy Jassy further indicates that similar to the commercial rollout of AWS, the company will likely deploy proprietary AI tools and features internally at AMZN, ahead of commercializing the products.
Although this approach might take a while to establish, we believe the path is more likely to reflect in a broader uptake of AI, across enterprises. AMZN will likely upsell AI solutions for typical use cases within organizations, with options to customize. Net-net, it appears that AI uptake will probably be a slog. There is no silver bullet.
Hardly A Sink Or Swim Situation
Based on financial results, it appears that MSFT’s AI platform is garnering interest. Over F4Q24, revenues associated with AI accounted for 8% of the 29% year-over-year growth attributed to the firm’s Azure cloud computing segment. In addition, MSFT’s GitHub Copilot accounted for 40% of GitHub’s year-over-year growth in FY24. Further, ~70,000 organizations have adopted GitHub Copilot. Moreover, on an annualized basis, the Microsoft 365 Copilot customer base expanded 60% over F4Q24. Furthermore, the number of companies with more than 10,000 Microsoft 365 Copilot seats doubled on a quarter-over-quarter basis. Finally, sales associated with Team’s Premium version, which includes Copilot, advanced 400% compared to F4Q23. Overall, although we view the customer uptake of MSFT’s AI products as encouraging, the outcome is far from dramatic, as the growth is building of a small absolute base. In addition, rapid early adoption of novel technology is typical.
MSFT is a large corporation, and AI presently represents a small fraction of its cloud computing business. During FY24, the company generated $245 billion in revenues, reflecting a growth of 16%, over the prior year. The Productivity and Business Process category generated $78 billion in sales, Intelligent Cloud (where the AI opportunity resides) $115 billion, and More Personal Computing $62 billion. 32% of MSFT’s revenues were attributed to its Productivity and Business Process segment, 43% to Intelligent Cloud, and 25% to More Personal Computing. Sales associated with the respective segments expanded by 12%, 20%, and 13%, compared to FY23. Gross margins, operating margins, and profit margins for FY24 came in at 70%, 45%, and 36%.
Clearly, although MSFT is firing on all cylinders, the prosperity is not based on AI. AI represents a future significant growth driver, not a current one. In our opinion, even if the AI opportunity fades relative to indicated potential, MSFT will continue to deliver.
Bottom Line
We’ve been skeptical about MSFT in the past. We did not appreciate how Bill Gates drove Internet Explorer and Microsoft Search Network into the ground, and how Steve Ballmer completely missed out on the mobile phenomenon. And, although we view MSFT’s current popularity in AI as serendipitous, based completely on its decision to partner with OpenAI, which fortunately began to deliver boatloads of AI goodies, we can’t fault MSFT in its efforts to take the AI ball and run with it.
The endeavor has us viewing the company through the arc of time. MSFT was founded in 1975, and although it could have delivered more profits for shareholders than it has, the firm has stood the test of time. MSFT makes mistakes and moves on to bigger and better things. It is certainly not a race horse, but it is clearly a workhorse.
AI might not work out (or it might), but MSFT as a company is likely to dust itself off and fight. After all, tomorrow is another day.
Analyst’s Disclosure: I/we have no stock, option or similar derivative position in any of the companies mentioned, and no plans to initiate any such positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Seeking Alpha’s Disclosure: Past performance is no guarantee of future results. No recommendation or advice is being given as to whether any investment is suitable for a particular investor. Any views or opinions expressed above may not reflect those of Seeking Alpha as a whole. Seeking Alpha is not a licensed securities dealer, broker or US investment adviser or investment bank. Our analysts are third party authors that include both professional investors and individual investors who may not be licensed or certified by any institute or regulatory body.