Alongside the software announcements, Microsoft has unveiled the Surface RTX Spark Dev Box, a compact AI developer workstation powered by NVIDIA’s RTX Spark superchip. Featuring a grace CPU, blackwell RTX GPU and 128GB of unified memory, the system delivers up to 1 petaflop of AI performance, enabling developers to run models with more than 120 billion parameters locally.

The launch highlights Microsoft’s broader push towards edge-AI computing, where increasing AI workloads can be executed directly on increasingly-capable devices, rather than in the cloud.

According to our Global Quarterly AI Laptop Tracker and Forecast, AI Advanced PC penetration is projected to reach approximately 70% of global shipments in 2030, up from 37% in 2025. The Surface RTX Spark Dev Box and Microsoft’s focus on the NPU moment reflects a broader strategy to strengthen Window’s role and opportunity in AI PC development. Rather than focusing solely on larger cloud models, Microsoft is betting that future AI workloads will be distributed across cloud infrastructure, GPUs, and increasingly capable NPUs embedded in PCs

The announcement reflects Microsoft’s broader push to create an end-to-end AI stack spanning silicon, operating systems, developer tools, and cloud services.

Microsoft unveiled Majorana 2, its next generation quantum processor promising more stable and durable qubits. The company claims the chip delivers 1,000 times greater qubit reliability than its predecessor Majorana 1, marking a transition from proving the underlying physics to advancing towards scalable engineering. Microsoft also updated its roadmap, targeting a commecially viable, fault-tolerant quantum computer by 2029 approximately two years earlier than previously expected.

Microsoft’s focus on Majorana 2 at Build 2026 relects its continued effort to bring quantum computing closer to mainstream enterprise conversations. While practical, large-scale quantum applications are still some distance away, the company is using Azure and its developer ecosystem to help organizations explore potential use cases and build familiarity with quantum technologies.

Microsoft launched seven new in-house AI models under the MAI (Microsoft AI) brand. Making its biggest step yet toward developing proprietary foundation models rather than relying primarily on OpenAI.

MAI-Thinking-1 is Microsoft’s first proprietary reasoning model and is designed to compete with leading models from OpenAI and Anthropic while keeping costs lower. The company says it performs well on a range of reasoning and software engineering tasks, suggesting that businesses may not have to choose between strong performance and affordability.

Microsoft AI has positioned itself as a contender in the global AI race by prioritizing economic viability and multimodal capabilities for enterprise-scale deployments rather than competing purely on model size.

This approach reflects a broader shift in the industry, where long-term adoption is increasingly determined not only by model intelligence, but also by the ability to deliver AI at a sustainable cost across text, speech, image, and coding applications.

The launch is part of Microsoft’s broader effort to build more of its AI capabilities in-house, while still giving customers access to models from OpenAI, Anthropic, and other providers through its Azure AI Foundry.

Counterpoint Research’s Global LLM Adoption Snapshot tracks Monthly Active Users (MAUs) across leading LLMs worldwide, giving a data-driven context to benchmark Microsoft’s MAI push against ChatGPT, Gemini, Claude and others.

Azure HorizonDB entered public preview at Build as a fully managed, postgreSQL compatible database designed specifically for AI-native and agent-driven applications. Microsoft positioned the service as a response to a growing challenge, traditional database architectures were not built to support autonomous agents that require low-latency access to consistent, real-time context across distributed environments.

As the foundational data layer for Foundry-based agent deployments, HorizonDB is intended to help enterprises build and scale AI applications that depend on persistent memory, coordination and state management.

Microsoft’s expanding security framework for Agentic AI and the general availability of the Agent 365 SDK, embeds compliance enforcement, access controls, governance policies, and observability directly into the agent development lifecycle, reflecting Microsoft’s view that enterprise AI security must be built into applications from the outset rather than added later.

Microsoft also extended its security-first strategy into software development with MDASH, a scanning harness that integrates Microsoft Defender and GitHub to identify and remediate vulnerabilities using AI. By embedding automated security analysis directly into development workflows, Microsoft aims to reduce risk throughout the software lifecycle while accelerating the adoption of AI-assisted coding and agent-driven applications.

From an industry perspective, the Agent 365 SDK and related security announcements reflect Microsoft’s efforts to address one of the key barriers to enterprise AI adoption: governance and security.

As organisations move from AI assistants to more autonomous agents, concerns around identity, access control, data protection, and oversight become increasingly important. By extending enterprise security and management capabilities to AI agents, Microsoft is positioning its platform to support large-scale agent deployments while giving organisations greater visibility and control.

Microsoft is executing on a deeply integrated platform strategy. Azure serves as the foundation cloud layer. While the MAI model family reduces third-party dependency. Copilot and Scout capture the productivity and workflow layer. GitHub owns developer mindshare. Project Solara aims for a new hardware category. And the Majorana program stakes a claim in the next decade of computing.

Microsoft’s AI revenue run rate surpassed $37 billion in its most recent quarter, growing 123% YoY. Azure cloud services grew 39% in the same period, with AI contributing an estimated 13% to 16% of that growth, indicating that AI is now responsible for roughly one-third of Azure’s revenue expansion.

Microsoft’s scope expanded from being an AI-enabled software company to becoming a full-stack AI platform provider. The biggest competitive gain was reducing its dependence on OpenAI through MAI models and Azure AI Foundry, while the biggest growth opportunity is the emergence of enterprise AI agents.

If agent adoption scales over the next three to five years, Build 2026 may be remembered as the point where Microsoft moved beyond copilots and began building the infrastructure for the enterprise AI economy.