Recent financial reports reveal decelerating revenue growth for leading cloud service providers, despite maintaining double-digit expansion. Industry perspectives are being shaped by DeepSeek's skepticism toward large hardware investments and an increasing focus on software and AI innovations. Additionally, uncertainty looms over potential new restrictions on AI infrastructure, including chips, that may be implemented by the Trump Administration.
Cloud momentum declines
Amazon Web Services (AWS) has seen its annual revenue growth slow from 30-40% in 2021 to 10-20% in recent years. Microsoft's Intelligent Cloud segment has maintained relative stability, though its growth rate has decreased from 20-30% to 10-20%. Similarly, Google Cloud, under Alphabet, has experienced a deceleration in annual revenue growth, dropping from 40-50% to 20-30%.
Infrastructure investments continue
Major cloud providers continue to increase their capital expenditures. Amazon CEO Andy Jassy has emphasized that AWS expands capex only when strong demand signals emerge. He describes AI as a "once-in-a-lifetime business opportunity" and views AWS's increased investment in the sector as a positive indicator for long-term growth.
Amazon announced capital investments of US$26.3 billion for the fourth quarter of 2024, with projections indicating this rate will likely remain steady in 2025. Similar to 2024, a significant portion will support growing technology infrastructure demand, primarily in artificial intelligence and cloud services. The remaining investments will go toward retail operations, including new stores and delivery infrastructure improvements.
Alphabet reported a fourth quarter of 2024 capex of US$14 billion, primarily invested in servers and data centers. CFO Anat Ashkenazi projects that capex will increase to US$75 billion in 2025, largely due to AI-related technical infrastructure investments. Of this amount, US$16 billion to US$18 billion is expected in the first quarter of 2025.
Microsoft announced a fiscal second quarter of 2025 capex of US$22.6 billion, including finance leases, meeting projections. CFO Amy Hood indicates that capex for the fiscal third and fourth quarters of 2025 will remain consistent with second-quarter levels as the company continues investing in AI-driven infrastructure.
DeepSeek disrupts AI landscape
DeepSeek's impressive AI model performance has captured attention across the AI supply chain, despite industry skepticism.
Arm and Qualcomm have shown particular enthusiasm for DeepSeek's advancements. While Alphabet acknowledges DeepSeek's capabilities, it maintains confidence in Gemini's technological edge. However, DeepSeek's emergence as a major AI competitor has raised questions about Microsoft's relationship with OpenAI.
During the latest earnings call, Microsoft reaffirmed its commitment to OpenAI while emphasizing its broader industry engagement. In response to analyst inquiries, CEO Satya Nadella confirmed the strong partnership, highlighting OpenAI's commitment to Microsoft's Azure cloud platform and the resulting financial benefits.
Nevertheless, Nadella emphasized software optimizations' role in reducing AI costs. He referenced DeepSeek and their collaboration with OpenAI to reduce GPT model pricing, explaining that powerful AI models must be cost-effective to drive demand.
According to Nadella, strategic hardware investment timing is crucial, as Moore's Law ensures technology improvements while optimization enhances efficiency. He advocates for continuous upgrades and modernization to balance monetization and training expenses.
Policy uncertainties under new administration
Beyond tariff policies raising industrial concerns and potentially disrupting global supply chains again, questions remain about whether the Trump Administration will implement additional AI infrastructure restrictions.
In January 2025, the Biden Administration introduced a new US AI diffusion policy and export controls. Under these rules, US-based companies and allies must become Universal Validated End Users (UVEUs) within a year by meeting security standards that reduce technology theft or diversion risks to China.
UVEUs face no restrictions on controlled AI chip exports. US-based UVEUs such as Microsoft, Alphabet, and Amazon must maintain at least 50% of their AI computing within the US, 75% within the US and allied nations, and no more than 7% in any other country, ensuring major data centers remain in trusted regions per US government standards.
The regulations require cloud providers to keep 75% of their AI computing resources within the US and allied nations, while resources in other countries cannot exceed 7% of total capacity. This policy could influence AI data center strategies for major companies including Microsoft, Amazon, and Alphabet.
US-based companies face fewer restrictions compared to non-US entities under this framework. The policy, effective January 13, 2025, allows exporters, re-exporters, and transferors until May 15, 2025, or January 25, 2026, to comply with changes regarding validated end users and exports of model weights under a new license exception. This timeline enables potential modifications by the Trump Administration.