July 30, 2026 —
Azure clears $100 billion as Microsoft’s AI buildout accelerates
Today’s AI Tools Radar brief is led by Microsoft’s clearest evidence yet that AI demand is translating into cloud scale: Azure crossed $100 billion in annual revenue as the company accelerated capacity, efficiency and Copilot adoption. Meta’s results show the other side of the race—capital spending is surging faster than earnings—while Arm, Intel, OpenAI, Google and AWS advanced the compute and software layers beneath the market.
Azure clears $100 billion as Microsoft’s AI buildout accelerates
Microsoft reported $90.0 billion in fiscal fourth-quarter revenue, up 18% year over year, while Azure and other cloud services grew 43%. For the full fiscal year, Azure revenue surpassed $100 billion and Microsoft Cloud reached $214 billion. Microsoft said it added 31 data centers during the quarter, brought another gigawatt of capacity online and cut the time needed to activate new GPUs in its largest regions by nearly half.
The company is also extracting more from that infrastructure: Copilot workload throughput has quadrupled since the start of the year, Microsoft 365 Copilot passed 30 million paid seats, and Foundry revenue more than doubled. Quarterly capital expenditures reached $41 billion as customer demand continued to exceed available Azure capacity.
Meta lifts AI infrastructure spending as profit contracts
Meta reported second-quarter revenue of $60.8 billion, up 28%, but net income fell 14% as costs expanded sharply. Capital expenditures, including finance-lease principal payments, reached $31.1 billion for the quarter, and Meta narrowed its full-year capital-spending forecast to $130–145 billion. The company said AI is already improving its core business while opening opportunities in agents, APIs, compute and enterprise software.
Arm opens fiscal 2027 with record first-quarter revenue
Arm said it delivered a record first quarter for total revenue as demand for its compute platform broadened across cloud, edge and physical AI. The result follows the company’s expansion from processor intellectual property into compute subsystems and its first production silicon for agentic workloads. Arm’s position inside hyperscaler CPUs and custom silicon makes its royalty and licensing trajectory an important read on non-GPU AI infrastructure.
OpenAI triples ARC-AGI-3 performance by changing the harness
OpenAI said two Responses API settings tripled GPT-5.6 Sol’s score on ARC-AGI-3 while using fewer output tokens. The changes retained reasoning state between turns and enabled context compaction, giving the model a more coherent and efficient way to interact with the benchmark’s unfamiliar puzzle environments. The result reframes part of agent performance as a systems problem involving memory, context and orchestration—not model weights alone.
Intel highlights U.S. packaging capacity for larger AI chiplets
Intel detailed how its Foveros stacking, EMIB bridges and EMIB-T technologies connect multiple chiplets into large AI packages at its New Mexico facilities. The company says current production can scale packages to eight times the conventional reticle limit, with a path beyond twelve times by 2028. The approach targets systems that need tightly integrated compute, memory and connectivity without relying on a single oversized die.
Google brings Lyria 3.5 to Flow Music
Google DeepMind launched Lyria 3.5 in Flow Music, emphasizing stronger musicality, vocals, lyrics and finer creative control for AI-assisted composition.
Bedrock AgentCore adds secretless agent authentication
AgentCore Identity now supports Private Key JWT, keeping signing keys in AWS KMS while recording agent token access through CloudTrail.
AWS publishes a governed MCP architecture for business intelligence
A new AgentCore reference design connects enterprise data through MCP servers while enforcing identity, Cedar policies, isolated runtimes and persistent memory.
Amazon Quick turns retention analysis into an agent workflow
AWS demonstrated a no-code retention pipeline that analyzes calls and satisfaction data, prioritizes at-risk customers and drafts personalized outreach.
Microsoft frames AI security as a systems problem
Microsoft urged organizations to pair AI-driven security analysis with layered controls, continuous monitoring, governance, validation and clear human accountability.