Themes by year
| Theme | 2026 | 2025 | 2024 |
|---|---|---|---|
| AgentsAI agents, agentic workflows, agent platforms and protocols such as MCP and A2A. | 17 | 23 | 8 |
| AI infrastructureChips, GPUs, accelerators, AI data centers and factories, networking and compute. | 8 | 12 | 7 |
| Data platformsDatabases, lakehouses, warehouses, streaming, analytics and data engineering. | 5 | 8 | 8 |
| ModelsNew frontier and foundation models and model families from the host. | 1 | 7 | 6 |
| AssistantsConsumer and workplace AI assistants: Siri, the Gemini app, Copilot, ChatGPT features. | 6 | 4 | 3 |
| AI codingAI for software development: coding agents, developer tools, code review. | 3 | 5 | 4 |
| Open sourceOpen-source software and communities: Kubernetes, Linux, Postgres and more. | 1 | 6 | 5 |
| Enterprise appsBusiness applications with AI: CRM, ERP, IT service, HR, marketing, creative suites. | 3 | 4 | 3 |
| DevicesPhones, PCs, wearables, glasses, headsets and other consumer hardware. | 4 | 3 | 2 |
| SecurityCybersecurity, identity and the security of AI systems. | 4 | 2 | 3 |
| GovernanceAI governance, trust, responsible AI, compliance and sovereignty. | 4 | 1 | 4 |
| RoboticsRobotics, physical AI, autonomous vehicles, simulation and digital twins. | 2 | 3 | 3 |
| PlatformsOperating systems and app platforms: iOS, Android, Windows, the web, design tools. | 2 | 2 | 4 |
| CloudCloud platform services, migration, hybrid and multicloud, serverless. | 1 | 3 | 1 |
| PricingPricing, packaging, licensing and free tiers as a headline. | 1 | 3 | · |
| Open modelsOpen-weight models and the ecosystems around them. | 2 | 1 | · |
| Keynotes on file | 23 | 33 | 23 |
A cell counts that year's keynotes tagged with the theme; darker means a larger share of that year's keynotes. Years with fewer than five keynotes on file, and keynotes still pending a summary, are left out of the grid.
What the themes mean
Agents: AI agents, agentic workflows, agent platforms and protocols such as MCP and A2A.
Assistants: Consumer and workplace AI assistants: Siri, the Gemini app, Copilot, ChatGPT features.
AI coding: AI for software development: coding agents, developer tools, code review.
Models: New frontier and foundation models and model families from the host.
Open models: Open-weight models and the ecosystems around them.
AI infrastructure: Chips, GPUs, accelerators, AI data centers and factories, networking and compute.
Data platforms: Databases, lakehouses, warehouses, streaming, analytics and data engineering.
Cloud: Cloud platform services, migration, hybrid and multicloud, serverless.
Devices: Phones, PCs, wearables, glasses, headsets and other consumer hardware.
Robotics: Robotics, physical AI, autonomous vehicles, simulation and digital twins.
Security: Cybersecurity, identity and the security of AI systems.
Governance: AI governance, trust, responsible AI, compliance and sovereignty.
Enterprise apps: Business applications with AI: CRM, ERP, IT service, HR, marketing, creative suites.
Open source: Open-source software and communities: Kubernetes, Linux, Postgres and more.
Platforms: Operating systems and app platforms: iOS, Android, Windows, the web, design tools.
Pricing: Pricing, packaging, licensing and free tiers as a headline.
Themes are picked from this fixed list only, 1 to 3 per keynote, most central first, from the transcript or the official recap.