Microsoft Corp (NASDAQ:MSFT) is seeing a sharp acceleration in enterprise rollouts of its Copilot artificial intelligence tools as the product matures and large customers move from gradual testing to company-wide deployments, according to Jefferies.
The brokerage, which reiterated its “Buy” rating and a $675 price target, said Microsoft executives signaled broadening momentum at a recent conference, with the company closing last year’s performance gap with OpenAI and now offering “much stronger ROI and enterprise readiness.”
Jefferies said Microsoft’s commercial sales chief told investors that once Copilot deployments hit a critical seat threshold, usage tends to inflect. The firm’s customer checks showed that many businesses are now shifting to enterprise-wide adoption.
The push into AI “agents” remains at a very early stage, however, with Microsoft’s commercial chief Judson Althoff saying the company is still “in the first half of the first inning.” The company recently introduced Agent 365 as a unified control plane for enterprise AI agents, designed to handle registry, governance and third-party integrations.
AI demand is also lifting traditional cloud workloads, Jefferies said, noting that each AI GPU task can require up to 10 supporting CPU virtual machines, straining Microsoft’s infrastructure. The company remains capacity-constrained and must keep expanding data centers to meet both AI and non-AI needs.
Microsoft continues to gain share in core cloud migrations thanks to its end-to-end platform, the report said, citing customer checks that show the company winning 75% to 85% of large SAP migrations over Amazon Web Services. About 70% of enterprise workloads remain on-premises, leaving substantial room for further cloud adoption.
Jefferies also highlighted Microsoft’s expanding “IQ layer,” which aims to unify enterprise data for AI applications and could provide a competitive edge over AI-native rivals that rely on fragmented data connectors.
The bank expects modest upside to Microsoft’s revenue forecasts for fiscal 2026 and 2027, though it models a 6% EPS shortfall versus consensus in 2027.