AWS, Microsoft, and Google Cloud reveal sharply different AI bets in Q3

AWS, Microsoft, and Google Cloud stake out divergent AI paths

Amazon Web Services, Microsoft Azure, and Google Cloud just delivered third-quarter earnings that revealed three competing visions for AI and cloud computing. AWS revenue jumped 39.3% to $46 billion. Microsoft Azure grew 48% to $33.2 billion. Google Cloud nearly doubled, surging 103% to $30.7 billion. Yet beneath Google’s cloud success, Alphabet’s core ad business faltered: Search revenue fell below 50% of total revenue for the first time, a sign the company’s financial engine is sputtering.

AWS pushes choice and custom silicon to widen its lead

AWS kept defying expectations, posting its sixth straight quarter of accelerating growth. Year-over-year revenue climbed $13 billion—almost what the entire division brought in during a single quarter five years ago. Operating income reached $17.8 billion at a 38.7% margin, while its backlog ballooned to $578 billion, more than double the $244 billion it reported at the end of 2025. CEO Andy Jassy told analysts growth could exceed 40% in both 2027 and 2028 and suggested AWS might cross the $1 trillion revenue mark within five years.

The bet is on choice. Bedrock now offers more than 100 AI models from Anthropic, OpenAI, Meta, Mistral, DeepSeek, Google, xAI, and Nvidia, letting enterprises swap models under one security umbrella. That model neutrality puts daylight between AWS and Microsoft’s OpenAI-centric approach or Google’s Gemini-first playbook. Already, Fortune 500 spending on Bedrock has risen nearly sevenfold year-over-year, and token volume is up 17-fold.

Cost savings are the other pillar. Trainium, Amazon’s inference chip, helped customers cut more than $2 billion in expenses during the quarter. Jassy also announced plans to sell Trainium outside AWS starting in late 2027, turning a cloud cost-center into a potential on-premises revenue stream that could challenge Nvidia.

Microsoft converts productivity suite into AI on-ramp

Microsoft turned its grip on productivity software into an AI on-ramp. Office, Teams, Windows, and now Copilot sit inside the same stack, creating a smooth path for enterprise adoption. The numbers tell the story: companies with more than 50,000 Copilot seats grew eightfold year-over-year. Capgemini, Hitachi, and the UK’s HMRC now run their operations on Azure, weaving Microsoft’s AI into their daily workflows.

The strategy is straightforward. Microsoft removes friction by offering AI inside tools customers already use. That predictability comes at a price—enterprises that want to experiment with other AI models or infrastructure may feel trapped. CFO Amy Hood acknowledged minor delays in third-party hardware deliveries but kept the focus on Microsoft’s balance sheet, which remains strong.

Google Cloud’s growth masks ad slowdown and cash burn

Google Cloud’s 103% revenue growth to $30.7 billion stood out in Alphabet’s earnings, but the rest of the business raised red flags. Search revenue, long the company’s profit engine, slipped below 50% of total revenue for the first time, growing only 12%, down from 19% in the first quarter. YouTube ad revenue missed expectations, climbing 9.2% to $11.21 billion instead of the $11.43 billion analysts had penciled in.

Google’s AI push centers on Gemini, its proprietary model. Tensor Processing Unit sales helped lift cloud revenue, yet its closed approach risks alienating enterprise customers who want more flexibility. Meanwhile, Alphabet’s spending spree grew: capital expenditure hit $51.4 billion, free cash flow turned negative $9.1 billion, and the company expects capex to top $300 billion in 2027. That cash burn raises questions about whether the growth is sustainable—or just expensive.

Oracle and IBM grow fast but lack the scale to challenge hyperscalers

Oracle’s cloud infrastructure revenue soared 121% to $7.4 billion, but its financials reveal a growth story built on debt and prepayments. Remaining performance obligations swelled to $664 billion, yet its debt load reached $125 billion. Free cash flow sank to negative $5.4 billion, and capital expenditure hit $28 billion in a single quarter—signs of a high-wire act.

IBM fared better but remains an also-ran. Revenue rose 5% to $17.1 billion, led by Red Hat, and free cash flow reached $7.8 billion year-to-date. Neither IBM nor Oracle, however, has the scale, flywheel, or ecosystem to rival AWS, Microsoft, or Google in the hyperscaler AI race.

Analysts favor AWS’s open approach in hyperscaler AI race

Analysts are picking sides. Zeus Kerravala of ZK Research called AWS the “long-term winner,” praising its open platform and model neutrality. He ranked Microsoft second, not because of its AI strategy, but because its installed base of productivity tools acts as a moat. Google, he argued, would finish a distant third—its closed Gemini strategy and weakening ad business leave it vulnerable.

Each company is making deliberate trade-offs. AWS gives enterprises flexibility and lower costs; Microsoft offers predictability and seamless integration; Google bets on rapid cloud growth but risks alienating customers who want openness. For CIOs, the choice comes down to priorities: model choice, integration speed, or the willingness to gamble on Google’s vision.

Key takeaways

• AWS’s open platform and custom silicon have pushed Fortune 500 spending on Bedrock up nearly sevenfold year-over-year, giving enterprises more choice and lower costs.
• Microsoft’s deep integration of Copilot into its productivity suite creates strong customer lock-in but may limit flexibility for companies that want to mix AI models or infrastructure.
• Google Cloud’s 103% growth looks impressive, but its reliance on a closed AI strategy and weakening ad revenue could hurt its long-term enterprise appeal.

Related on Lazyfounder

• Google
• OpenAI
• Nvidia
• Oracle
• More business news