Microsoft places huge cap-ex bets on datacenters for cloud and AI
CFO says paying customers expected to flood in from 2024
Microsoft is betting big on generative AI by spending many billions of dollars more on building datacenter capacity in anticipation of a rapid uptick in demand from paying customers, expected to start in anger from calendar 2024.
Despite tightening its belt on R&D dollars for the first time since 2016, capital expenditures in Q4 of Microsoft's fiscal 2023 ended 30 June were $10.7 billion, up from $7.8 billion in the prior three months.
Amy Hood, CFO at Microsoft, told analysts on a conference call the money is to "support cloud demand, including investment in AI infrastructure." This includes new datacenters plus more "CPUs and GPUs and networking equipment."
This elevated spend will continue well into the new financial year. "I think the real focus here is being able to be aggressive in meeting the demand curve," Hood said on a conference call in response to a question about the pace of adoption of generative AI.
Github Copilot – a coding assistant developed by GitHub and OpenAI – was released in October 2021. Individual developers are charged $10 a month or $100 for a year's subscription, and the Business version costs $19 per month per user. Github said in February that 1.2 million developers used the technical preview as of September 2022.
The only Copilot product to have gone GA, GitHub Copilot is not without its controversies, including questions about the training of the AI which uses volumes of publicly available source code, drawing on these materials to respond to prompts.
Microsoft is sprinkling Copilot across much of its portfolio from Windows, to 365, Dynamics and more. In all cases the use is designed to save time and unlock productivity gains. Microsoft is to charge $30 per user per month for Copilot under an E3, E5, Business Standard and Business Premium licenses.
"The pitch of a dollar a day to significantly increase the productivity of an employee might work," Rob Sanfilippo, analyst at Directions on Microsoft said last week.
Microsoft started to talk up its new Copilot-plus products in March, as the generative LLM bandwagon-jumping intensified. Microsoft 365 Copilot remains in private preview with 600 invited but paying customers, according to Directions on Microsoft. There is not yet a release date on when the product - which will be integrated into the user interface for Word, Excel, PowerPoint, Loop and Teams – will be used live production environments.
The analyst says the next product to roll off the GA conveyor belt is Dynamics 365. It won't be an extra cost to punters with certain premium software licenses and will be an add-on for others, "most likely Professional licenses."
Bing Chat Enterprise was released in preview last week for Microsoft 365 E3, E5, Business Premium and Business Standard license holders, and will be available to them at no additional cost. A standalone plan will be $5 a month.
Back at this week's conference call, Microsoft CFO Hood said investments in datacenters and AI is because the "demand signal is quite strong. It remains strong.
"I'm thrilled with the products announcements we've made, I'm thrilled with them moving to paid preview and then moving to GA. They absolutely are expansive in terms of addressable market. They reach new budget pools."
She added: "As you know we're focused on executing against that. And then revenue is an outcome… the demand signal requires the capital expense."
Hood expects the AI services to "pull along Azure," as customers use Microsoft products running on Azure and then use the public cloud's AI tools to build their own generative AI, as credit rating agency Moody's did recently.
Microsoft, along with AWS and the other major hyperscalers already account for 37 percent of the world's datacenter capacity, according to Synergy Research Group (SRG) – and that capacity is going to almost double over the next half decade.
John Dinsdale, chief analyst and research director at SRG, told us via an emailed comment the opportunities for companies that want to "lead the charge in cloud services are - buty so are the investments required.
"You need to have a network of hypserscale data centers that straddles the world. Amazon, Google and Microsoft each has 150-200 hyperscale data centers in operation, and each has another 80-100 in the pipeline.
You need to fund that scale and the constant technology changes that are driving innovation, hence the billions of dollars invested by each of market leaders every quarter. You need to be a company that has credibility and the right brand in all regions of the world. And you need total management commitment and a strategy that is built around long term vision rather than short term gains. There will be late arrivals, but they'll be picking up crumbs from the masters' tables," he added.
Over at Google, cap-ex investment was $13 billion in the first half of calendar 2023 versus $16.6 billion a year earlier. Ruth Porat, ex-CFO and newly crowned President and Chief Investment Officer, said part of this is related to "delays in certain datacenter construction projects."
"We expect elevated levels of investment in our technical infrastructure increasing through the back half of 2023 and continuing to grow in 2024. The primary driver is to support the opportunities we see in AI across Alphabet, including investments in GPUs and proprietary TPUs as well as datacenter capacity," she told analysts on the conference call this week.
- 1 in 4 Brits are playing with generative AI, and some take its word as gospel
- AI mentioned 175 times during Microsoft's Q4 earnings call
- OpenAI pauses Bing search feature over paywall bypass abilities
- Artificial General Intelligence remains a distant dream despite LLM boom
"And as we continue to see the pace of innovation accelerate, we just want to make sure we're positioned to address the opportunity across Alphabet."
Scott Kessler, Third Bridge global sector lead for tech media and telecommunications, estimates Google spent "upwards of $200 billion on AI investments over the past decade."
"Much of that isn't necessarily appreciated by users and investors," he told Reuters.
In one sense, cost is becoming an ever higher barrier to entry into the world of technology, something that big fish like Microsoft and Google probably don't hate. ®