Microsoft (US:MSFT), Amazon (US:AMZN) and Alphabet (US:GOOGL) may be the market’s darlings, but right now the only thing investors really care about is their cloud computing performance. Microsoft and Alphabet both reported their recent third-quarter results on the same day, and both beat consensus earnings expectations. However, Microsoft’s share price rose 4 per cent and Alphabet’s fell 9 per cent.
The reason? Microsoft’s earnings beat was built on the back of cloud computing while Alphabet’s was due to a recovery in its advertising market.
Microsoft’s cloud computing division, Azure, grew 29 per cent year on year in the three months to September, exceeding market expectations of 26 per cent. This extra growth came from artificial intelligence (AI), with chief financial officer Amy Hood confirming that roughly three percentage points came from AI services. “While the trends from [the] prior quarter continued, growth was ahead of expectations, primarily driven by increased graphics processing unit (GPU) capacity and better-than-expected GPU utilisation of our AI services,” said Hood.
In the three months to September, Alphabet’s revenue increased 11 per cent to $76.6bn (£63bn), which was 1 per cent ahead of market expectations. Meanwhile, earnings per share rose 46 per cent to $1.55, more than 6 per cent ahead of analyst predictions. The problem was Google Cloud. It ‘only’ grew 22 per cent year on year, a slowdown from 29 per cent last quarter and below what analysts were hoping for.
On Thursday, Amazon’s share price was up just 5 per cent despite its earnings smashing broker expectations. Between, Monday and Friday of that week its share price was basically flat. This evident disappointment was linked again to its cloud division, which made $23.1bn – slightly behind the $23.2bn forecast. This 12 per cent year-on-year growth was the same as last quarter. Thankfully for investors, earnings per share of $0.96 was more than 60 per cent ahead of expectations, thanks to a surprisingly strong performance from its retail and advertising businesses.
On the earnings call, chief executive Andrew Jassy was keen to ease fears that the company had fallen behind in the AI race. He mentioned the custom Trainium chip it has designed, the recent deal with OpenAI rival Anthropic and its ‘AI-as-a-service’ platform called Bedrock. He name-checked customers building AI apps on the platform, including Adidas, Booking.com, Relx (REL), Merck (DE:MERK) and United Airlines (US:UAL).
In total, the phrase ‘generative AI’ was used 31 times on the call as management laid on the point thickly. “It’s a gigantic new generative AI opportunity which I believe with be tens of billions of dollars of revenue for AWS over the next several years,” said Jassy
Microsoft, Alphabet and Amazon’s share prices have risen this year, not because of their legacy businesses, but because of their cloud computing potential. All have spent billions on capital expenditure to build out the physical infrastructure needed for widespread AI adoption.
Microsoft’s big advantage
The problem for Amazon and Alphabet is that they have legacy businesses which are mostly unrelated to cloud computing. Amazon is dragging along one of the world’s biggest retail businesses, which struggles to break even. AWS has an operating margin of 30 per cent but it makes up just 16 per cent of total revenue.
Similarly, 78 per cent of Alphabet’s revenue comes from advertising through YouTube and Google Search, while just 19 per cent comes from the cloud. The upside is that advertising is a profitable business, with operating margins of around 35 per cent, but it is not a growth industry. Digital advertising is now dominant and won’t grow much faster than the wider economy.
The advantage Microsoft has is the crossover with its legacy business. Microsoft has more than 300mn Office 365 users, most of whom use their software ‘on-premise’. In other words, they downloaded the software onto their own computers and servers, rather than accessing it through Microsoft Azure.
Now, through its strategic investment in OpenAI, Microsoft has moved ahead of its competitors. It has already launched Github Copilot, an AI assistant which speeds up coding and has more than 1mn paid users and 37,000 organisations subscribed. It is now rolling out an Office 365 Copilot, which will add generative AI to its suite of Office 365 products, including Excel, Word and Powerpoint.
Analysts like the fact Microsoft already has a large base of customers to whom it can upsell its AI products. “Microsoft is still in the early innings of a major opportunity with AI and we think its cloud product could reach 50 per cent of its install base,” said Dan Ives from Wedbush Securities.
To use these AI products, customers must be on the cloud. Software was traditionally run ‘on-premise’. However, the computing power needed to run AI-enabled software means it must be done on the huge cloud servers. For Microsoft, it can go to its existing customers, sell them its new AI software and then transition them to the cloud.
IT services company Bytes Technology (BYIT), Microsoft’s largest reselling partner in the UK, is starting to benefit from the AI interest. In the six months to August, its gross invoiced income was up 38 per cent year on year. “We have been inundated with queries about AI and how to prepare to onboard this technology and what kind of infrastructure they need,” chief executive Neil Murphy told Investors’ Chronicle last week.
Cloud exposure, without the rest
If investors just want cloud computing exposure, there are software infrastructure companies that aren’t held back by any slow-growth legacy businesses.
Cloud monitoring company Datadog (US:DDOG) allows customers to monitor their technology stack and collect data on cloud usage, network traffic and memory. It’s rather like having a smart meter for cloud computing costs.
In the three months to June, Datadog’s revenue rose 25 per cent to $509mn. Meanwhile, the number of customers with more than $100,000 of annual recurring revenue rose 24 per cent to 2,990. It has just announced a large language model (LLM) “observability solution”. In other words, if a company builds a chatbot using its proprietary data, as Relx has done with its AI Legal Assistant, it can use Datadog’s software to monitor its functionality and detect issues.
The other infrastructure beneficiary is cloud data storage and analysis company Snowflake (US:SNOW). Its software allows customers to collate and analyse their data in the cloud, and with the growth of generative AI the returns on being able to organise and process data grow substantially.
Its new product, Snowpark, allows customers to train AI models on the underlying Snowflake data, rather than ingesting it into the platform, cleaning it, then exporting it to train and run proprietary models. Snowflake’s $2.1bn of revenue in the year to January was up 248 per cent on two years ago.
The promise for these companies lies in the flywheel effect. As AI improves, software solutions to problems arising will become cheaper. Research from GitHub showed that developers are 55 per cent more efficient using Github Copilot. Of course, these exact numbers can be debated, but AI is undeniably making development easier. This means more software solutions to problems and more data to be analysed, and more cloud infrastructure services.
Soon, all telephone communications will be done in the cloud. In 2025, BT (BT.) is shutting down its public switched telephone network (PSTN), which will force all communication onto fibre-optic cables. This is part of the investment case for Gamma Communications (GAMA), whose software allows businesses to make calls over the internet using their existing telephone number. This will be in the cloud, where conversations will be stored as data and probably used to build AI call centre products. Read more on Gamma here.
As the economy stumbles, it is rare to find growth in places that aren’t linked to cloud computing, data storage and generative AI. With billions of dolllars invested, it has to work. Even slight stumbles towards this promised future can be hugely damaging.