Cursor’s $50 billion valuation and Copilot’s usage shift

Cursor negotiates a $2 billion fundraising round at a $50 billion valuation while GitHub Copilot users face unpredictable costs due to a new AI Credits model. Engineering leaders must navigate the divergence between Cursor’s request-based pricing and Copilot’s…

Cursor's $50 billion valuation and Copilot's usage shift

Cursor negotiates a $2 billion fundraising round at a valuation of over $50 billion. This valuation does not include the new investment. The company forecasts annualized revenue of more than $6 billion by the end of 2026. In February, Cursor reached $2 billion in annualized revenue. Enterprise customers now provide 60% of Cursor revenue, a significant increase from 25% eighteen months ago. This growth occurs while GitHub Copilot users face unpredictable token costs as Microsoft shifts toward a usage-based model.

GitHub Copilot moved to an AI Credits model on June 1, 2026. The base subscription price remains unchanged. Copilot Enterprise costs $39 per seat per month. This price includes $39 in monthly AI Credits. These credits pool across the organization. One AI Credit equals $0.01 USD. Developers consume these credits when they use premium models like Claude Opus or o3. Standard autocomplete on basic models remains cheap. An agentic session for 30 minutes on a complex codebase can burn a large portion of a single user’s monthly allotment.

The math behind Copilot credits

GitHub Copilot Enterprise usage depends on token consumption across input, output, and cached tokens. The seat price is fixed at $39, but the actual bill varies. The massive divergence between the aggressive growth projections for Cursor and the unpredictable token consumption models used by GitHub Copilot creates a significant and lasting headache for engineering leaders who must reconcile high-speed agentic workflows with fixed monthly procurement budgets.

Agentic workflows consume 5x to 20x more tokens than standard autocomplete. Copilot agent mode, automated PR review, and workspace agents drive this consumption. A single ambitious agent session that uses 4 million input tokens and 400,000 output tokens costs $18. This session uses 1,800 AI Credits. This uses nearly half of one Enterprise seat’s $39 allotment. One engineer can burn half a month’s allotment in an afternoon.

Plan Price Included
Copilot Business $19/seat/mo AI Credits allotment, SSO, content exclusions
Copilot Enterprise $39/seat/mo Higher credits, SAML SSO, audit logs, IP indemnity

Total cost of ownership at scale

Engineering leaders in organizations with 200 to 1,000 engineers make procurement decisions based on usage profiles. Light usage involves standard autocomplete and minimal agentic use. Moderate usage includes multi-turn chat and occasional agent sessions. Heavy agentic usage involves regular use of agents and automated PR review bots.

For 200 engineers, Copilot Enterprise costs approximately $7,800 per month for light usage and $12,000 to $18,000 for heavy agentic use. Cursor Business costs approximately $8,000 for light usage and $10,000 to $15,000 for heavy agentic use. At heavy usage, Copilot’s credit model creates unpredictability because overages depend on token consumption. Cursor’s request-based model is more enumerable because a fast request is a fast request.

Usage Profile Copilot Enterprise (200 Eng) Cursor Business (200 Eng) Copilot Enterprise (500 Eng) Cursor Business (500 Eng)
Light ~$7,800/mo ~$8,000/mo ~$19,500/mo ~$20,000/mo
Moderate ~$8,500 – $10,000/mo ~$8,000 – $9,500/mo ~$21,000 – $25,000/mo ~$20,000 – $24,000/mo
Heavy Agentic $12,000 – $18,000+/mo $10,000 – $15,000+/mo $30,000 – $45,000+/mo $25,000 – $37,000+/mo

Governance and compliance requirements

Cursor Business lacks SSO, SCIM provisioning, and audit logs. These features are standard requirements for any organization with 200 or more engineers. Companies that need these compliance tools must purchase Cursor Enterprise. Cursor does not publish Enterprise pricing. This requires a custom contract and a direct sales process. Copilot Business and Enterprise include these features.

Copilot also provides IP indemnity at the Enterprise tier. This is a meaningful differentiator for organizations that worry about copyright exposure. GitHub also integrates deeply with Microsoft’s identity stack. This makes Copilot a natural choice for teams that already use Azure AD or Microsoft 365.

Feature Copilot Business Copilot Enterprise Cursor Business Cursor Enterprise
SSO / SAML Yes Yes No Yes
SCIM provisioning Yes Yes No Yes
Audit logs Basic Full No Yes
Centralized billing Yes Yes Yes Yes
Usage analytics Basic Enhanced Yes Yes
Policy management Yes Yes No Yes

Cursor request-based pricing mechanics

Cursor uses a request-based billing model. Each fast request is a call to a high-quality model. Developers use these for autocomplete, chat, or agentic sessions. Cursor Pro costs $20 per month and includes 500 fast requests. Cursor Business costs $40 per seat per month and includes 1,000 fast requests.

A developer running Cursor Agent for a 2-hour refactoring session might consume 40 to 80 fast requests. If a developer hits the cap, they fall back to slow requests. Cursor also offers a Premium seat for $120 per month. This seat provides 5 times the usage of the Standard seat for only 3 times the cost. The Premium seat is designed for developers who run agents all day.

Plan Price Included
Cursor Pro $20/mo 500 fast requests/mo
Cursor Business $40/seat/mo 1,000 fast requests/mo
Cursor Enterprise Custom Custom request pools

The September credit cliff

September 1, 2026, marks the end of promotional credit allotments for GitHub Copilot. GitHub gave Business and Enterprise customers extra credits in June, July, and August 2026. Business seats received 3,000 credits. Enterprise seats received 7,000 credits. Standard allotments are 1,900 for Business and 3,900 for Enterprise.

On September 1, Enterprise pools shrink by 44% and Business pools shrink by 37%. The seat price does not change. Most Enterprise admins have not seen their real bill yet. June, July, and August usage fell on a subsidized pool. Organizations that adopted agent workflows heavily over the summer will see higher invoices in October.

Tool fragmentation in the enterprise

59% of developers use three or more AI coding tools simultaneously. Many enterprise organizations standardize on Copilot but still have teams using Cursor, Claude Code, or Codeium. These teams often expense tools individually outside of central procurement. This behavior causes budget leakage. A $39 Copilot seat and a $40 Cursor seat create a $79 baseline per seat before any usage overages.

Fragmentation also creates compliance gaps. If Cursor runs outside of IT visibility, audit logs on the Copilot side do not capture the full picture. Different teams develop different muscle memory on different tools. This makes knowledge sharing harder across the engineering department. You should identify which tools your team actually uses before you sign a new contract.

Decision logic for engineering leaders

The choice between tools depends on the specific needs of the engineering team. Choose GitHub Copilot Enterprise if the organization standardizes on GitHub Enterprise. It is the best option if the team needs IP indemnity or deep integration with GitHub Issues and PRs. It is also the correct choice for teams that want to avoid switching editors.

Choose Cursor if the developers want an AI-native editor experience. Cursor is the better option for teams that heavily use agentic workflows and value model flexibility. You should pick Cursor if your team needs per-developer usage analytics without waiting for GitHub to update its dashboard.

Tool Best For Main Advantage
GitHub Copilot Teams on GitHub Deep integration and compatibility
Cursor AI-native workflows Advanced agentic capabilities
Claude Code Large tasks Terminal-based autonomy

Will Microsoft lower its seat prices to counteract the rise of agentic-specific IDEs? Most organizations will find that the decision is not about the base seat price. The decision is about how the team manages token consumption and agentic workflows. Standardize on Copilot for compatibility or move to Cursor for agentic power.

airtrain.ai
airtrain.ai

The airtrain.ai newsroom covers AI research, models and the tools built on them.

More on this topic

Stay ahead of AI

Get the week's most important AI stories delivered to your inbox every Monday.

No spam. Unsubscribe anytime.

More Stories