Generative artificial intelligence (AI) is a hot ticket these days, and some of the world's largest tech companies are building their futures around AI strategies. At the same time, game-changing AI can pose serious business risks. Tech titan Microsoft (MSFT 1.57%) is not afraid to share its AI-related risk analysis with the public, and you should take a look at management's concerns before making any AI investments -- in Microsoft or other tech stocks.
NASDAQ: MSFT
Key Data Points
How Microsoft talks about AI risks
A recent Motley Fool research report reviewed how five of the "Magnificent Seven" companies discuss AI topics on their earnings calls in the first three quarters of calendar year 2024.
None of the tech giants spent a whole lot of conference call time on AI risks, but Microsoft did stand out as a leading analyst of risk factors.
Amazon (AMZN 0.87%) devoted only 4% of its AI mentions to risk factors, mostly focused on the high costs of setting up powerful AI services. Apple (AAPL 2.15%) gave even less space to risk discussions, with just 3% of the mentions centered on strict regulations slowing down the rollout of Apple's AI tools.
Microsoft took a different approach and directed 10% of its AI mentions toward risk analysis. More broadly, the company related almost all of its AI talk to the Copilot AI assistant tool and the Azure cloud-computing platform, and the risk discussion stayed within those guardrails to develop a theme of extreme demand making it tough to satisfy every order for Azure and Copilot AI services.
Risk 1: Building AI infrastructure is a massive and expensive challenge
Microsoft CEO Satya Nadella is not surprised by the soaring demand for generative AI services, but real-world limitations still make it difficult to capitalize on that demand.
"We have run into obviously lots of external constraints because this demand all showed up pretty fast, right?" Nadella said on the Q1 2025 earnings call in October 2024. "We ran into a set of constraints, which are everything because [data centers] (DCs) don't get built overnight. So there is DCs. There is power. And so that's sort of been the short-term constraint."
Building lots of data center capacity and securing large-scale electric power feeds is a slow and expensive process, but Microsoft is building assets with massive long-term value.
Risk 2: It's hard to keep up with surging AI demand
Despite its enormous business scale and generous infrastructure investments, Microsoft still finds it hard to keep up with the soaring AI demand.
- In the April 2024 call, CFO Amy Hood noted that "near-term AI demand is a bit higher than our available capacity."
- Three months later, Hood explained that Azure's AI demand "remained higher than our available capacity."
- I just showed you the October call, where Nadella highlighted the "constraints" holding back Microsoft's AI business. On the same call, Hood stated that "demand continues to be higher than our available capacity."
- Jumping ahead to last month's second-quarter call, Hood simply repeated her demand analysis from October. There's no stopping the inflow of AI-related orders for Azure's cloud-based computing capacity.
Risk 3: DeepSeek and other up-and-coming competitors
More recently, the Chinese DeepSeek engine emerged as a threat to the current AI leaders. A small team with a modest budget for computing hardware can generate AI results comparable to the big-budget leaders. Of course, DeepSeek came up often in the second-quarter call, earning more than 8% of Microsoft's total AI discussion.
Nadella said that the AI market is developing much like the old-school computing industry did. New innovations drive down costs and improve the quality of computing services, and the entire market is always adapting to a steady stream of game-changing shifts.
"Obviously, now that all gets commoditized and it's going to get broadly used," Nadella said. "And the big beneficiaries of any software cycle like that is the customers."
So DeepSeek presents new risks to Microsoft's AI business but it's also a normal and healthy part of this emerging market. We're about to find out what happens when deep-pocketed tech giants apply huge data centers and computing budgets to the same technology that worked wonders at DeepSeek's limited scale.
In the end, AI tools from the phone in your pocket to massive-scale cloud solutions will benefit from unexpected challenges such as the DeepSeek approach. And that's why you won't be stuck with today's AI quality in the long term, just like you're not using Pentium processors or Windows 3.1 in 2025.
You can't stop progress. You can only slow it down. But why would you want to?
