Microsoft Copilot added 5.6 million mobile users in the U.S. between March and June 2025, a 175% surge that crushed ChatGPT’s 3.9 million adds and 17.9% growth rate over the same period, according to exclusive Comscore data shared with ADWEEK. The numbers reveal an AI assistant market where enterprise distribution and deep product integration are accelerating adoption faster than raw consumer appeal, even as ChatGPT retains a commanding absolute lead with 25.4 million mobile users.

The latest figures from Comscore’s consumer AI tool usage dataset—tracking 117 tools across desktop and mobile—show that total mobile web and app usage of AI assistants reached about 73.4 million users in the March–June window, while desktop usage dipped to roughly 78.4 million. Copilot’s mobile user base jumped to an estimated 8.8 million, while Google’s Gemini grew 68% to 14.3 million, partly driven by preloading on Pixel phones. The dataset is drawn from a digital panel of roughly a quarter-million devices.

“The productivity element of Copilot, and the convenience from an enterprise perspective—given that it’s integrated into Microsoft products—is driving a lot of the adoption,” Brian Pugh, chief product officer at Comscore, told ADWEEK. That enterprise-first strategy is the core differentiator: Copilot is not just a standalone app but is woven into Microsoft 365, Windows, Edge, and Azure, creating multiple low-friction entry points for business users.

Why Enterprise Distribution Changes the Game

For IT teams, the path to Copilot adoption is paved with familiar admin tools: it can be rolled out via existing Microsoft 365 licenses or as an add-on, with single sign-on, directory integration, and compliance controls already in place. That reduces the behavioral friction that hamstrings many third-party AI tools inside large organizations. Instead of asking workers to install yet another app and create new credentials, Copilot appears inside Word, Outlook, Teams, and other daily productivity tools.

Comscore analysts point specifically to this productivity narrative and enterprise convenience as a major adoption vector. When a company standardizes on a managed AI assistant, the structural advantages are formidable: IT can enforce data loss prevention, audit usage, and control which models process sensitive information. For regulated industries, that governance layer is often the deciding factor that tips the balance away from consumer-oriented alternatives.

Bundling and Device Deals Accelerate the Shift

Beyond enterprise licensing, hardware bundling and carrier partnerships are reshaping user acquisition economics. Google’s Gemini benefits from being preloaded on Pixel phones, a tactic that helped push its mobile user count to 14.3 million. The Comscore data underscores that channel strategy—not just raw model capability—can move millions of users quickly. Microsoft has its own levers: Copilot is integrated into Surface devices, and the company is striking deals with OEMs and mobile carriers to surface the assistant more prominently. Such moves create a virtuous cycle where distribution begets habit, which begets stickiness.

Comscore’s findings also reveal that more than 85% of top AI assistant users stick to a single platform, suggesting these tools are becoming habitual workflow components rather than one-off experiments. Once Copilot becomes the default assistant inside a company’s productivity stack, the switching costs for users and admins are nontrivial.

Scale vs. Velocity: The ChatGPT Paradox

ChatGPT remains the undeniable mass-market leader. With 25.4 million mobile users and a developer ecosystem that includes plugins, GPTs, and APIs, it commands the largest absolute audience. But its 17.9% growth—while still strong—pales next to Copilot’s 175% surge. The divergence illustrates a market moving at two speeds: ChatGPT grows steadily from a large base, while Copilot accelerates faster from a smaller one by leveraging enterprise muscle.

That dynamic has implications. For consumers and developers, ChatGPT’s broad reach ensures it remains a critical part of the AI fabric, with OpenAI’s rapid revenue scaling financing continued R&D. For enterprises, however, Copilot’s integration advantage makes it the default choice for many procurement cycles. The right read: Copilot is not overtaking ChatGPT in total users anytime soon, but it is capturing the high-value business segment that management consultancies call “the AI enterprise wedge.”

The Commercial Engine: Azure and OpenAI Revenues

The adoption figures translate directly into cloud consumption, subscriptions, and enterprise deals. Microsoft recently disclosed that Azure surpassed $75 billion in annual revenue, growing roughly 34% year-over-year, with executives attributing much of the lift to AI workloads. That revenue funds massive capex for data centers, networking, and specialized inferencing hardware. Meanwhile, OpenAI reported hitting roughly $10 billion in annual recurring revenue from ChatGPT subscriptions and API sales as of June 2025, per multiple outlets. That milestone confirms that both consumer and developer monetization vectors are accelerating—and explains why cloud providers and device vendors are structuring commercial incentives heavily around AI.

These financials create a flywheel: enterprises buy Copilot seats and Azure services, which fuels infrastructure investment and product features; consumer subscriptions to ChatGPT fund model R&D and API availability; device deals embed assistants at the OS level. The winners in this cycle will be those that can bundle AI deeply into existing ecosystems without alienating users or regulators.

What’s Driving Copilot Adoption

Several factors give Copilot its current momentum:

  • Deep Microsoft ecosystem integration: Presence in Word, Excel, Outlook, Teams, and Windows reduces the behavior change needed to adopt an AI assistant.
  • Enterprise governance and compliance: Admin controls, data loss prevention, and auditing features align with corporate security requirements.
  • Channel reach: Microsoft’s vast network of resellers, MSPs, and system integrators accelerates deployment at scale.
  • Bundling economics: Copilot can be folded into Microsoft 365 subscriptions and device plans, lowering the incremental cost for IT buyers.

For Windows users, these gains promise faster productivity improvements—from smarter search in Edge to on-device assistance for common tasks. IT leaders get a way to formalize AI governance with defined data boundaries and controlled connectors, rather than letting “shadow AI” usage proliferate.

Risks, Measurement Caveats, and Regulatory Shadows

No dataset is perfect. Comscore’s panel-based measurement captures active usage, but deduplication rules, session definitions, and sampling can skew comparisons. App download counts (reported by firms like Sensor Tower) often differ from monthly active users. High percentage growth from a low base can look dramatic while still yielding modest absolute user counts. And OpenAI’s $10B ARR figure, while publicly stated, excludes certain licensing and one-time deals, so independent auditors have not published a full breakdown.

Product risks also loom:

  • Fragmentation: Multiple Copilot versions (consumer, Microsoft 365, GitHub) can confuse legacy users.
  • Feature parity: Some users still find that consumer assistants like ChatGPT offer better document-handling or conversational features, making internal adoption uneven.
  • Reliability and hallucinations: Enterprise customers require predictable outputs and provenance; model errors can undermine trust unless mitigated with retrieval-augmented workflows and human review.

Regulatory scrutiny is intensifying. Bundling an assistant across a dominant productivity suite raises antitrust questions in the EU and U.S. if competitors claim the practice unfairly disadvantages them. Heavy integration into a single vendor’s stack also creates lock-in that can be expensive to unwind. Privacy and data governance remain persistent concerns, even with Microsoft’s compliance features.

Tactical Guidance for Enterprises and IT Teams

Enterprises evaluating Copilot should take a measured approach:

  1. Identify high-ROI use cases first—summarizing email threads, drafting templates, automating form-heavy tasks—before rolling out broadly.
  2. Pilot with governance controls enabled—data loss prevention, output logging, and human validation for sensitive workflows.
  3. Measure business outcomes, not just user counts. Track time saved, error rates, and compliance metrics.
  4. Consider hybrid architectures: Use Azure-hosted tooling for regulated data while leveraging selective third-party APIs where features are superior but governance is maintained.
  5. Negotiate commercial terms with future portability in mind—request data exportability and termination clauses that preserve long-term choice.

Impact on Windows and the Broader AI Ecosystem

For Windows users, Copilot’s gains are generally good news. Tighter integration means faster access to AI-powered features in the operating system and native apps. Microsoft’s cloud scale and per-seat licensing let IT roll out the assistant with familiar management tools. But the consolidation also raises strategic questions: as Copilot becomes the default, independent AI startups may find it harder to break into enterprise accounts, and consumer choice could narrow.

For competitors, the lesson is stark: distribution channels and device partnerships matter as much as raw model capability. Expect more preloads, more bundling, and more vertically integrated AI offerings from the major platform owners. Policymakers will watch closely for signs that bundling stifles innovation or harms consumers.

What to Watch Next

  • Quarterly Comscore releases will reveal whether Copilot’s Q2 mobile momentum persists across cohorts and geographies.
  • Microsoft’s product roadmap for deeper on-device capabilities, agentic workflows, and Edge/Windows integration will determine if the assistant becomes indispensable or one option among many.
  • OpenAI’s commercial disclosures and product expansions—including enterprise GPTs, APIs, and possible hardware tie-ups—will shape developer strategy, especially given the company’s $10B ARR base.
  • Regulatory developments in major markets (EU, U.S.) could influence bundling practices, ad productization inside assistants, and preload agreements.

The Comscore data captures a pivotal moment: Copilot is accelerating on mobile by leveraging Microsoft’s enterprise channels, while ChatGPT holds a far larger audience and strong developer momentum. For enterprises and Windows users, the practical imperative is to evaluate assistants on governance, integration, and measurable outcomes—not headline growth. At the same time, the industry must grapple with the structural implications of bundling, consolidation, and the future of productivity software.