Foundation Models Cross the Chasm as Enterprise Wallets Shift to Anthropic

Generative AI applications exited the experimental phase in 2025, with Menlo Ventures documenting that enterprise spending on the category reached $37 billion – triple the prior year – while Anthropic captured 40% of enterprise foundation model spend compared to OpenAI’s 27%.

The market concentration is stark: OpenAI, Gemini, and Anthropic together command 89% of enterprise wallet share, according to Andreessen Horowitz. Anthropic now earns 40% of enterprise LLM spend, up from 24% last year and 12% in 2023, while OpenAI lost nearly half of its enterprise share, falling to 27% from 50% in 2023. On the consumer side, Claude’s app reached 11 million daily users in 2026, while Anthropic’s Claude reached over one million new signups every single day in early March, according to Android Headlines. The adoption velocity confirms that AI applications have moved from proof-of-concept to production infrastructure.

Developer Economics Reshape the Stack

Anthropic’s ascent has been driven by its remarkably durable dominance in the coding market, where it now commands an estimated 54% market share, compared to 21% for OpenAI, up from 42% just six months ago, driven in large part by the popularity of Claude Code. Departmental AI spending hit $7.3 billion in 2025, up 4.1x year over year, with coding the clear standout at $4.0 billion (55% of departmental AI spend), according to Menlo Ventures.

The economics favor winner-take-most concentration. Within one month of Claude 4’s release, Claude 4 Sonnet captured 45% of Anthropic users while Sonnet 3.5 share decreased from 83% to 16% – even as individual models drop 10x in price, builders don’t capture savings by using older models; they just move en masse to the best performing one. This pattern creates a trap for smaller AI companies: they must match frontier model performance while competing on price against providers who benefit from massive scale and can afford to commoditize inference.

There are now at least 10 products generating over $1 billion in ARR and 50 products generating over $100 million in ARR, led by the model APIs powering applications (Anthropic, OpenAI, Google), but increasingly distributed across departmental solutions in coding, sales, customer support, HR, and verticals from healthcare and legal to the creator economy. The revenue concentration at the infrastructure layer means application developers face a structural disadvantage: they pay the toll to access Foundation Models while competing with those same providers’ application-layer products.

The SaaS Displacement Question

Enterprise Software markets are repricing disruption risk. Since the start of 2026, ETFs for public software companies have fallen by 30 percent, erasing all the gains since the launch of ChatGPT, with companies like Salesforce, Adobe, Intuit, ServiceNow, and Veeva down 25 to 30 percent in a matter of weeks, reports Andreessen Horowitz.

The displacement thesis rests on three mechanisms. First, AI agents in 2026 are replacing traditional enterprise SaaS platforms by executing workflows autonomously rather than presenting dashboards, with enterprises cutting licensing costs, reducing human dependency, and shifting from seat-based pricing to compute-based AI orchestration models – autonomous AI systems expected to handle 60–80% of routine enterprise workflows by 2027 according to research trends from IBM, Gartner, McKinsey, Microsoft, and ServiceNow cited by Gammatek Solutions.

Second, 35% of respondents said they have replaced functionality of at least one SaaS tool with a custom build, and 78 percent expect to build more of their own tools in 2026, according to a Retool survey reported by Newsweek. Third, seat count compression reduces software revenue even when the software itself survives. When AI can do the work of multiple humans, you need fewer humans; when you need fewer humans, you need fewer seats – if 10 AI agents can do the work of 100 sales reps, you don’t need 100 Salesforce seats anymore, you need 10, a 90% reduction in seat revenue for the same work output, writes Jason Lemkin at SaaStr.

But replacement risk is uneven. Harvey and Hebbia are building finance and legal collaboration spaces that connect service providers and clients on a single system: the more people and agents who use these platforms, the more valuable the platforms become – insofar as AI makes the network more powerful, we should expect to see AI make these network effects more powerful than they were before. Workflow complexity and switching costs matter more than feature velocity.

Consolidation Math for Startups

Enterprises are testing multiple tools for a single-use case, and there’s an explosion of startups focused on certain buying centers like go-to-market, where it’s extremely hard to discern differentiation even during proof of concepts – as enterprises see real proof points from AI, they’ll cut out some of the experimentation budget, rationalize overlapping tools and deploy that savings into the AI technologies that have delivered, Andrew Ferguson of Databricks Ventures told TechCrunch.

The capital environment reinforces concentration. Global venture investment totaled $189 billion in February – the largest startup funding month on record – although 83% of capital raised went to just three companies, including OpenAI, which raised $110 billion, also in the largest round ever raised by a private, venture-backed company, and Anthropic, which raised $30 billion, marking the third-largest venture round on record, reports Crunchbase. Last year’s venture funding disproportionately went to a select group of companies – OpenAI, Scale AI, Anthropic, Project Prometheus and xAI each raised more than $5 billion in 2025, with those five companies raising $84 billion, or 20% of all venture funding last year.

While AI continues to capture a disproportionate share of venture dollars, fewer startups are being funded overall but those that are secure significantly larger rounds – early 2026 has already seen at least 17 U.S. AI startups raise $100M+ rounds, signaling that Venture Capital is behaving less like exploratory capital and more like institutional underwriting, according to VentureBurn.

The critical distinction is between infrastructure and applications. Capital is consolidating around entities that control infrastructure layers, not applications built on top of them. For the thousands of AI startups that raised seed rounds in 2023-2024 on the promise of building vertical AI solutions, the 2026 funding environment presents a binary outcome: demonstrate durable competitive advantage through proprietary data, network effects, or regulatory moats – or face a down round as enterprise budgets consolidate to fewer vendors.

What to Watch

Three data points will clarify whether AI apps are building durable businesses or renting growth from foundation model providers. First, enterprise retention curves by spring earnings season – if AI-native companies show 120%+ net dollar retention, they’ve captured budgets rather than borrowed them. Second, developer platform churn rates for startups building on Claude and OpenAI APIs – rising API costs and direct competition from foundation model providers will surface as margin compression in Q2 reports. Third, Series B pricing for 2023-vintage AI application companies – valuation resets will confirm whether proprietary data and workflow embedding create defensibility or whether investors are repricing these businesses as distribution channels for foundation models.

Menlo Ventures’ report predicts that long-horizon agents will drive the next major evolution of the enterprise AI stack. The companies that survive the transition from experimentation to production will be those that own the customer relationship, control proprietary data, or embed so deeply into operations that replacement cost exceeds switching benefit. For everyone else, the question isn’t whether AI will eat software – it’s who captures the revenue when it does.