Hyperscalers, AI growth & capex

Artificial intelligence (AI) remains one of the most important growth themes in technology, but the latest results from Amazon, Alphabet, Microsoft and Meta Platforms show that investors need to look beyond the headline excitement. The broad message is encouraging – demand for AI services is growing, businesses are moving from testing to real-world deployment, and the largest technology platforms are investing at enormous scale to meet that demand. At the same time, the cost of staying in the race is rising quickly, and the market is becoming more selective about which companies are turning AI spending into clear financial returns.

Moving from promise to practice

One of the clearest takeaways is that AI demand is no longer just a future story. It is already showing up in cloud growth, software adoption, advertising performance and customer usage trends. Across the group, management teams described strong demand for computing capacity, large increases in AI-related workloads and growing evidence that AI is becoming embedded in day-to-day products and services. In simple terms, the AI boom is moving from promise to practice.

That said, the environment is not uniform. Microsoft, Alphabet and Amazon are seeing especially strong momentum in cloud and enterprise AI, where customers are paying for infrastructure, tools and applications. Meta is benefiting from AI too, but in a different way. Its biggest gains are still showing up inside its advertising engine and user engagement, while investors remain less convinced that its very large AI spending will create a meaningful new non-advertising profit stream in the near term.

Multiple ways to win for Microsoft 

Microsoft remains one of the clearest examples of AI translating into visible commercial progress. Azure growth stayed strong, and management said demand continues to exceed supply. In its third-quarter update that covered the three month to the end of March, the company also pointed to an AI business that has already surpassed a $37bn annual revenue run rate, up 123% year over year. That is an eye-catching figure because it suggests AI is not just helping Microsoft protect its existing businesses; it is becoming a material business in its own right. 

Just as important, Microsoft is seeing traction across several layers of the stack, a layered architecture of hardware, software, data, and services that work together to create, train, and run artificial intelligence applications. Azure, its cloud computing business, remains a key growth engine, Foundry, its enterprise AI platform, is seeing strong customer activity, Cosmos DB, a cloud based database service, is benefiting from AI application workloads, and its AI assistant Microsoft 365 Copilot now has 20 million paid seats. Microsoft appears to be monetising AI through both infrastructure and software, which gives it multiple ways to win as adoption grows.

AI tools reinforcing Alphabet’s current business

Alphabet is also building a compelling AI story, though its model looks somewhat different from its peers. Alphabet is emphasising its “full-stack” approach, with its own frontier models through Gemini, its own silicon through Tensor Processing Units (TPUs), its infrastructure, and its broad distribution across consumer and enterprise products. That foundation appears to be supporting momentum across the business. 

Search queries reached all-time highs in the first quarter, as AI Overviews (AI‑generated summaries shown at the top of Google search results) and AI Mode (AI first search function) expanded, helping ease concerns that generative AI would weaken traditional search. At the same time, Google Cloud delivered a major acceleration, with revenue up 63% year over year along with improved profitability. A backlog of $462bn points to strong business demand, while growth in usage of its AI platform Gemini Enterprise suggests customers are increasingly paying for AI tools rather than merely experimenting with them. Taken together, the message is that AI is reinforcing both Alphabet’s consumer ecosystem and its enterprise offering.

Amazon: a meaningful transition

Amazon’s AI outlook is tied closely to AWS (Amazon Web Services), and here too the signs are encouraging. AWS delivered its fastest growth in 15 quarters, reaching a $150bn annualised revenue run rate. Management also said its AI revenue run rate now exceeds $15bn, while its custom silicon business has grown into a $20bn-plus annualised operation. That matters because Amazon is not only selling AI services; it is also building some of the hardware and infrastructure needed to deliver them. 

Amazon’s Trainium chips, Bedrock platform, a fully managed service for building generative AI applications, and managed agent offerings all point to a company that wants to serve customers from the underlying compute layer up to the finished AI application. There was also evidence that customer behaviour is changing. Bedrock processed more tokens in the quarter than in all prior years combined, which suggests generative AI usage is moving beyond pilot projects and into production. For investors, that is a meaningful shift because production workloads tend to be stickier and more valuable than early testing.

Meta needs to bridge investment-revenue gap

Meta is the most complicated case of the four. Its business remains highly profitable, and AI is clearly improving performance in core advertising. Ad impressions and pricing both improved, helping support strong revenue growth in the Family of Apps business. In other words, AI is already making Meta better at matching content and advertising with users, which is valuable because advertising is still the company’s main earnings engine. However, the market seems less willing to reward Meta simply for spending heavily on AI infrastructure without clearer evidence of incremental revenue outside ads. 

Full-year capital spending guidance moved higher, but near-term revenue guidance did not move up in the same way. That gap appears to be one reason investors reacted more cautiously. Meta’s newest large language model effort shows ambition, but the current debate is whether the company can create enough new value from AI products to justify the pace of spending.

Spending levels a quandary for investors 

All four companies are spending aggressively, and that has become one of the defining features of the current AI cycle. Alphabet raised its capital expenditure outlook and signalled another significant increase next year. Microsoft expects capital spending to keep climbing. Amazon’s capital expenditure has risen sharply, pressuring free cash flow. Meta also raised its full-year capital spending range. For long-term investors, this is a reminder that AI is not a low-cost opportunity. The leaders are building data centres, buying chips, securing memory and storage, and redesigning their product stacks around AI. The reward could be substantial, but the upfront cost is equally real.

Supply remains another important part of the story. Several companies indicated that demand is brushing up against limits in compute, memory or broader infrastructure. Microsoft said customer demand still exceeds supply. Alphabet said it remains compute constrained. Amazon pointed to higher component costs in memory and storage, even as it uses its scale to manage supply. These comments matter because they suggest the current pace of AI growth may be partly held back by available capacity, not by weak customer interest. Put differently, the problem today looks more like keeping up with demand than trying to create it.

The bottom line

For investors, the outlook therefore looks positive but more nuanced than the early AI excitement suggested. Microsoft appears to have one of the strongest combinations of demand visibility and monetisation. Alphabet is showing that AI can strengthen search while also accelerating cloud growth. Amazon looks increasingly well positioned as customers bring AI workloads into production and as its custom silicon strategy improves economics. Meta is proving that AI can deepen engagement and improve ad performance, but it still has more to prove when it comes to turning AI investment into a broader revenue story.

The big picture is that AI growth across the hyperscalers remains healthy, and in some cases is accelerating. Demand is broad, investment is massive, and the technology is becoming more embedded in both consumer and enterprise behaviour. The main debate has shifted. It is no longer whether AI matters. It clearly does. The question now is which companies can convert that importance into durable, profitable growth.

That distinction will likely shape performance from here. Markets may continue to reward the companies that can show rising AI demand and clearer monetisation, while being more cautious on those where spending is running ahead of visible returns. Even so, the overall direction is hard to miss. AI is becoming a core layer of the digital economy, and these four companies are helping build it.