In a major development for the AI sector, the enterprise LLM API market is witnessing a change of guard — Anthropic has reportedly surpassed OpenAI in usage among enterprise customers. This article delves into the latest market data, the factors behind the shift, implications for businesses and the competitive landscape, and what we should watch next.
What’s happened: The market share shift
According to a mid-year report by Menlo Ventures, enterprise spending on large language model (LLM) APIs has ballooned to US $8.4 billion in just six months into 2025.
Key figures from the report:
- OpenAI held about 50% of enterprise usage share by end of 2023.
- As of mid-2025, OpenAI’s share has fallen to around 25%.
- Anthropic now leads with about 32% of enterprise usage share.
- Other players: Google DeepMind/Google’s LLMs ~20%, Meta Platforms’s Llama ~9%.
So the headline: Anthropic has overtaken OpenAI in the enterprise LLM API market share (at least according to this survey).
Why this change occurred
Several factors appear to have contributed to Anthropic’s rise and OpenAI’s relative decline in this enterprise segment:
1. Enterprise-centric features & trust
Anthropic has emphasised enterprise readiness — data governance, compliance, integration with enterprise workflows — which is resonating with business buyers seeking not just powerful models but trustworthy, scalable solutions
2. Model performance and developer tooling
Anthropic’s recent model releases (e.g., Claude Sonnet 3.5, Claude 3.7, Claude 4) have advanced capabilities in reasoning and code generation. Also reports suggest Anthropic captured ~42% share in code-generation among developers vs OpenAI’s ~21%.
3. Shift from experimentation to production
The report notes that enterprise users are increasingly deploying LLMs for inference (production workloads) rather than just training or proofs-of-concept. With usage moving from experimentation to operational scale, durable vendor relationships and production readiness matter more. GlobeNewswire
4. Vendor switching is limited
Only ~11% of teams reported switching vendors in the past year; most (66%) upgraded the same vendor’s model. This means gaining share is harder — and Anthropic did it.
Implications for the industry & businesses
For OpenAI
- OpenAI’s dominance in consumer awareness (via ChatGPT) does not directly translate to enterprise dominance.
- Losing ground in enterprise APIs may pressure them to focus more on enterprise-specific features, governance, integrations.
- Must defend its position or risk further share erosion.
For Anthropic
- Becoming the leader in enterprise LLM APIs strengthens its positioning as an enterprise-first AI vendor.
- This can lead to more enterprise deals, higher revenue; but also greater expectations around support, SLAs, enterprise features.
- Must scale operations, infrastructure, compliance capabilities robustly.
For other players (Google, Meta, etc)
- Google rising at ~20% suggests the market is not a two-horse race any more.
- Meta’s 9% shows niche/secondary players still matter.
- The field may fragment around use cases, domains (e.g., code generation, agents, reasoning) rather than a single monopoly.
For enterprise buyers
- Having more choice: you’re not locked into OpenAI.
- Need to evaluate enterprise models based on actual deployment, governance, integration, not just brand.
- Watch model maturity, vendor lock-in, cost per token, support.
- Consider multi-vendor strategies rather than single provider dependency.
What to watch / next steps
- Sustainability of market share: The data is based on usage share in production workloads — will Anthropic maintain or extend this lead?
- Pricing & cost models: As usage scales, token costs, licensing, enterprise agreements will matter.
- Enterprise features & compliance: Data residency, privacy, integration with enterprise tooling will become differentiators.
- Open-source models or new entrants: While the report shows open‐source share falling (from 19% to ~13%), disruption could still occur.
- Regional/global adoption: Much of the enterprise AI story is US/large enterprises — what about India, APAC, MID-East?
- Use-case specialization: Code generation, agents, domain specific models (legal, healthcare) may drive next shifts.
- Vendor switching / churn: Even though switching is low now, if service issues or pricing changes occur, vendor churn may accelerate.
Conclusion
The enterprise LLM API market is evolving fast, and the current data indicates a notable shift: Anthropic has overtaken OpenAI as the top vendor in production enterprise usage, capturing ~32% share vs OpenAI’s ~25%. This signals that in enterprise deployments, considerations like trust, integration, performance and enterprise readiness are now as important (or more) than brand or hype.
As the market matures, the vendor landscape may become more competitive and specialized. For businesses, this means more vendor options — but also more complexity in making the right choice. OpenAI will need to adapt to defend its position, while Anthropic will need to scale and deliver robust enterprise-grade offerings to keep the lead.
