Hyperscalers - AI investment enters its next phase

Artificial Intelligence is entering a new chapter. Over the past two years, investors have focused on the enormous sums being invested in AI infrastructure. Today, the discussion is beginning to shift. Rather than asking how much the technology giants are spending, investors are increasingly asking how those investments will generate attractive returns.

The latest results from Amazon, Alphabet, Microsoft and Meta Platforms suggest that demand for AI remains exceptionally strong. Each company continues to invest heavily in data centres, chips and networking equipment, yet there is growing evidence that AI is moving beyond experimentation and into large-scale commercial use. At the same time, management teams are becoming more confident in explaining how these investments will be monetised over many years.

AI demand continues to outstrip supply

One of the clearest themes across all four companies is that customer demand for AI infrastructure continues to exceed available capacity.

Microsoft reported Azure revenue growth of 43% year-on-year and expects this to accelerate to 45% in the coming quarter. Azure has now surpassed a $100bn annual revenue run rate, while Microsoft added 31 data centres and one gigawatt of capacity during the quarter, with plans to double capacity over the next two years. The company stressed that accelerating growth reflects stronger customer demand rather than easing supply constraints.

Amazon is seeing a similar trend. AWS revenue accelerated to 37% year-on-year, reaching a $169bn annualised revenue run rate. Management highlighted that generative AI workloads are increasingly moving from the training phase into production, driving demand not only for AI computing but also for storage and databases that support these applications. AWS's backlog has grown to $496bn, supported by multi-year commitments from customers including Anthropic and OpenAI.

Alphabet also delivered exceptionally strong cloud growth, with Google Cloud revenue increasing 82% year-on-year. Enterprise adoption of Gemini Enterprise continues to expand rapidly, while nearly 90% of Fortune 100 companies are now using the platform. Even after significantly increasing infrastructure spending, management acknowledged that customer demand still exceeds internal capacity.

The consistent message across all three cloud providers is that demand remains stronger than supply, reinforcing confidence that AI investment is being driven by genuine customer adoption rather than speculative spending.

Capital expenditure remains high – but with improving visibility

The scale of investment required to support AI remains unprecedented.

Amazon has increased expected 2026 capital expenditure to around $220bn. Alphabet raised its annual capex guidance to $195-205bn, while Microsoft expects capital expenditure to continue growing during 2027 after spending $41bn in the latest quarter alone. Meta also increased the lower end of its annual capex guidance to $130-145bn.

These figures naturally raise questions about free cash flow and investment returns. Alphabet provides perhaps the clearest example, with free cash flow turning negative during the quarter as infrastructure spending exceeded operating cash flow. Management nevertheless emphasised that investment will continue while expected returns remain attractive.

Importantly, executives increasingly appear comfortable discussing the economics behind these investments.

Amazon explained that datacentre investment is inherently front-loaded. Infrastructure spending begins years before servers generate revenue, creating short-term pressure on free cash flow. However, management noted that server and networking equipment typically reaches breakeven in just under three years, while most AI capacity is contracted for at least five years. This combination provides confidence that attractive returns should emerge as utilisation increases.

Microsoft also highlighted that a growing proportion of investment is directed towards shorter-lived assets such as 

CPUs and GPUs rather than permanent infrastructure. This provides greater flexibility should demand eventually moderate, while the company's broad customer base supports high utilisation levels.

AI monetisation is becoming more visible

Perhaps the biggest change in this earnings season is that AI is increasingly generating meaningful revenue rather than remaining purely an investment story.

Microsoft continues to expand monetisation across multiple products. Microsoft 365 Copilot now has more than 30 million paid seats, GitHub Copilot has reached 50 million users, and Azure AI Foundry serves 100,000 customers. The company is also expanding "seat plus consumption" pricing across several product lines, allowing revenue to grow as customers use more AI services.

Amazon highlighted that both its AI revenue run rate and its custom silicon business have now exceeded $25bn annually, with both growing at triple-digit rates. At the same time, AI is supporting other parts of the business by improving advertising performance and operational efficiency across retail operations.

Alphabet's monetisation strategy spans both cloud computing and consumer products. Management noted encouraging monetisation from AI Overviews within Search while Google Cloud continues benefiting from enterprise AI deployments. Growth in Google One AI subscriptions also demonstrates that consumers are increasingly willing to pay for AI-enhanced services.

Meta's AI strategy is slightly different. While AI continues to improve advertising performance, management also highlighted significant longer-term opportunities in business agents to manage customer queries, product selection, pricing and payment, and in enterprise model APIs. These represent potential future revenue streams beyond digital advertising that management believes are not yet fully reflected in market expectations

Furthermore, Meta Compute, the recently launched cloud computing initiative will give management added flexibility in monetising their capex spend should some of these newer opportunities disappoint.

Return on invested capital remains a key debate

The largest question facing investors is whether these enormous investments will ultimately generate attractive returns.

Amazon addressed this issue directly, arguing that AI economics resemble the early development of AWS itself. While investment initially weighs on free cash flow, long-term customer contracts and rising utilisation should support attractive returns on invested capital over time.

Alphabet also expressed confidence in future returns, pointing to its integrated approach across software, customdesigned chips and cloud infrastructure. This allows the company to optimise hardware specifically for AI workloads while monetising AI through existing businesses including Search, YouTube, Cloud and subscriptions.

Microsoft focused on improving AI economics through lower inference costs, greater model efficiency and flexible architecture that allows customers to select the most appropriate AI models for different workloads. As AI becomes more efficient, customer adoption should broaden while supporting attractive returns for Microsoft itself.

Collectively, these comments suggest that management teams are increasingly focused not simply on building AI capacity but on demonstrating disciplined capital allocation and long-term profitability.

Risks remain

Despite the encouraging trends, several risks remain.

The largest near-term risk is that infrastructure investment continues to run ahead of monetisation. If enterprise AI adoption slows or customer demand disappoints, the current level of capital expenditure could place sustained pressure on cash flow and profitability.

Competition is also intensifying across AI infrastructure, cloud computing and consumer AI applications. Regulatory scrutiny remains elevated, particularly for Alphabet, Amazon and Meta, while macroeconomic uncertainty could influence both enterprise technology spending and advertising demand.

Outlook

The latest earnings reports suggest AI is entering a more mature phase of development. Infrastructure investment remains exceptionally high, but customer adoption is accelerating across cloud computing, enterprise software, search, advertising and productivity applications.

The conversation is also evolving. Investors are no longer focused solely on the size of capital expenditure. Instead, attention is increasingly turning towards utilisation rates, monetisation strategies and long-term returns on invested capital.

Across Amazon, Alphabet, Microsoft and Meta, management teams appear increasingly confident that today's investment will support sustainable growth over many years. While free cash flow may remain under pressure in the near term, the combination of rising customer demand, expanding AI products and improving commercial models provides growing evidence that AI is becoming a meaningful driver of long-term earnings growth rather than simply an expensive technology build-out.