Microsoft's AI ambitions spur investor debate, with questions abound

Seeking Alpha
2024.10.05 08:00
portai
I'm PortAI, I can summarize articles.

Microsoft's significant investments in AI, particularly in OpenAI, have raised investor concerns about monetization and ROI. Analyst Keith Weiss highlights that while Microsoft is seen as a leader in generative AI, investor patience is waning as shares lag behind the S&P 500. Weiss notes that revenue realization from AI capabilities will take time, with Azure AI and Microsoft 365 Copilot being key areas. Additionally, questions about capex returns, GPU availability, and operating margins persist, but Microsoft remains confident in meeting demand and maintaining low operating expenses.

Microsoft (NASDAQ:MSFT) has enormous artificial intelligence ambitions, spending billions of dollars investing in OpenAI and infusing AI into its own products and services, seemingly nonstop. While that ambition isn't slowing down, the questions investors have about its AI-related spending are also rising.

At the top of investors' collective mind is the monetization potential of generative AI, along with the return on investment in GPU spending and the environmental impacts of rising data center demand, Morgan Stanley analyst Keith Weiss wrote in a recent report.

Regarding generative AI revenue, investor patience "appears to be wearing thin," Weiss pointed out, as Microsoft shares have trailed the S&P 500 and Nasdaq year-to-date.

"As a perceived leader in the GenAI opportunity, investors look to Microsoft for indications on the proper time horizons for seeing revenues start to flow," Weiss wrote in a note to clients. Weiss has an Overweight rating and a $508 price target on Microsoft.

After speaking with investor relations executives, the takeaway was that Microsoft is confident, but it will take time to realize revenue from all of its new generative AI capabilities. "Importantly, while the company sees the macro as 'pretty stable for the past 6 quarters', they remain confident the demand for productivity themes holds up even in a weaker spending environment," Weiss wrote.

The issue of timing is likely best viewed in two bunches: Azure AI and Microsoft 365 Copilot, Weiss said. For Azure, AI-related solutions may have ended fiscal 2024 with $5B in annual run rate revenue. And the company still sees this as supply constrained, given how many customers are using the tools and the confidence that Microsoft management has in usage accelerating.

"With clear line of sight to data center capacity additions coming online in FY2H25, contracts in hand representing strong demand for GenAI solutions enabled by that capacity, and the denominator of the acceleration equation now somewhat smaller post the re-segmentation management remains confident in an acceleration which 'mathematically' should be a bit steeper now," Weiss wrote.

For Microsoft 365 Copilot, usage is the key to driving value, Weiss said.

"While still early in the product cycle, the M365 Copilot SKU is ramping faster than any prior SKU in the productivity suite, with [Vice President of Investor Relations Brett Iversen] stating, it 'feels like we are on track', as customers are now coming back to buying more, the solution is innovating quickly (as seen in the recent Wave 2 announcements), and the company is getting better at deploying and driving adoption," Weiss wrote.

Return on capex spend

The next issue is the return of investment on capex spending, which rose 75% year-over-year to $55.7B, including capital leases. In fiscal 2025, that's expected to rise another 41% to reach $78.4B.

The spending question may need to be split into two, Weiss posited, with one focused on the gross margin potential of GenAI related solutions, and the other on future model training costs.

Microsoft executives repeated the stance that was given on the most recent earnings call, that it's building to meet customer demand for generative AI functions.

Regarding gross margins, Iverson said he has "more confidence in the margin profile of these solutions" while adding the capital intensity of generative AI may not be higher than the cloud in the longer-term.

And while there are concerns about large language model training costs approaching $100B, Microsoft management said that figure is not anything they have ever spoken about and emphasized there needs to be "clear demand indications supporting the total investment behind the solutions – both training costs and the run-time costs of the solutions."

Forward growth

The next big question Microsoft investors have is constraints on future growth — particularly the availability of GPUs, along with the ability to power data centers and whether the company can meet its environmental targets.

Microsoft said while there are a number of engineering and procurement challenges that it faces building out its capabilities at scale, it does not see any part "which appears unsolvable," citing the recent deal with Constellation Energy (CEG).

Regarding GPU access, Microsoft management gave "little indication" on how the mix between first-party and third-party will evolve, with "importance of choice" and AMD's (AMD) offerings often brought up in the conversation.

Operating margins

The last big question investors have is the direction of operating margins.

For Microsoft's part, the company is confident that it can keep operating expenses low enough to offset capex spending, Weiss said.

"The strategy continues to be one of strong prioritization of spending initiatives, with management noting the rapid rise in the 15-year asset capex investment in FY24 making this a particularly difficult year to fully offset gross margin impacts," Weiss wrote. "Management noted opex growth in FY25 is not going to be as low as FY24. However, this is the first year internal AI initiatives are discreetly in the forward planning cycle, so Microsoft's ability to benefit from 'drinking their own champagne' is largely on the come."