Investors should prioritize Anthropic over OpenAI for enterprise exposure, as it has overtaken GPT in business adoption and is successfully expanding into application layers like design. While "Legacy" SaaS is under pressure, incumbents like Figma, HubSpot, and Google (GOOGL) remain high-conviction holds because their "sticky" seat-based revenue and bundled distribution are resisting the AI disruption. Look for specialized, model-agnostic tools like Cursor and OpenRouter that are gaining market share by helping businesses reduce costs through efficient task routing. A high-growth emerging opportunity exists in Answer Engine Optimization (AEO), with companies like Profund leading the shift from traditional SEO to AI-driven search visibility. Avoid over-weighting "pure" model providers that rely solely on token revenue, and instead focus on niche winners like Perplexity that offer superior search interfaces.

Anthropic (CLAUDE)

• Anthropic has recently overtaken OpenAI as the most popular AI model used by businesses, according to Ramp data. • The company is seeing significant "vertical growth" and is becoming a major competitor to established SaaS players by offering its own application layers (e.g., Claude Design). • Business adoption is driven by high product velocity and the ability to reach users where they are, though it faces competition from specialized tools like Cursor.

Takeaways

• Bullish Sentiment: Anthropic is currently winning the "model war" in terms of business spend growth, suggesting it is the current preferred choice for enterprise-grade reliability or performance. • Competitive Advantage: Its move into design (competing with Figma) shows that model labs are moving "up the stack" into applications, though incumbents with existing user bases remain sticky.

OpenAI (GPT)

• While previously the undisputed leader, OpenAI has slipped to the number two spot in terms of business adoption rates on the Ramp platform. • 80% of its business revenue is currently token-based (usage-based), which creates a conflict of interest: the company has no incentive to help users "auto-route" to cheaper models.

Takeaways

• Market Shift: The "first-mover" advantage is fading as businesses become more cost-conscious and willing to switch models based on performance and price. • Risk Factor: High dependency on token revenue makes them vulnerable to "efficiency" tools (like OpenRouter or Cursor) that help businesses spend less on expensive frontier models.

The "SaaSpocalypse" Theme

• The narrative that AI will "wipe out" Software-as-a-Service (SaaS) companies is currently not supported by data. • Seat-based pricing remains dominant (65-75% of spend), and businesses are not yet broadly switching to "agentic" or token-based payment models for their software. • Traditional SaaS companies like Figma and HubSpot are showing strong performance and growth despite new AI competitors.

Takeaways

• Investment Insight: Don't count out "Legacy" SaaS yet. Business spend is "sticky," and incumbents are successfully integrating AI features rather than being replaced by them. • Actionable Strategy: Look for SaaS companies that are "AI-native" but solve specific workflows (e.g., Adio in the CRM space) rather than just general model providers.

Cursor

• Cursor is highlighted as a major success story, effectively unseating GitHub Copilot (Microsoft) for many developers. • It provides a "bull case" for specialized AI applications because it can "auto-route" tasks to the most efficient model, saving users money and improving experience in a way that model labs (like OpenAI) won't.

Takeaways

• Disruption Potential: Small, nimble AI-native applications can disrupt massive incumbents (like Microsoft/GitHub) by focusing on a superior, model-agnostic user experience.

Perplexity AI

• Despite being a "pure play" model company that some analysts consider "at risk," Perplexity is one of the fastest-growing vendors on the Ramp platform. • It is succeeding by offering products and search capabilities that the major model labs have not yet prioritized.

Takeaways

• Niche Dominance: There is significant investment opportunity in companies that build specialized interfaces on top of models, even if they don't own the underlying "frontier" model.

Emerging Sector: Answer Engine Optimization (AEO)

• A new category of software is emerging to replace or augment SEO. • AEO helps firms track how they are being recommended within AI models (like ChatGPT or Claude). • Profund is mentioned as a specific company in this space growing extremely quickly.

Takeaways

• New Investment Frontier: As search shifts from Google to AI models, the entire "marketing tech" stack will be rebuilt. This is a high-growth area for early-stage investment.

Infrastructure & Cost Management (OpenRouter)

• Businesses are becoming increasingly "cost-conscious." Token costs for high-intensity users have increased 13x in the last year, which is unsustainable. • OpenRouter is seeing growth (3% of AI spend) as it allows businesses to route tasks to cheaper, open-source, or more efficient models.

Takeaways

• Efficiency Play: As AI spend matures, the "winners" will be the platforms that help companies reduce their AI bill by using the right model for the right task (e.g., using a small model for a summary and a frontier model for complex logic).

Google (GOOGL)

• Google's Gemini is described as "underrated" in business spend data. • Much of its adoption is hidden because it is bundled "for free" within Google Workspace.

Takeaways

• Distribution Advantage: Google has a massive "moat" because it already sits on the desktops of millions of workers. Even if it isn't winning the "paid API" race, its integration into daily workflows is a long-term strength.

xAI (Grok)

• xAI is a late entrant with roughly 3% adoption among businesses on the platform. • It has not yet seen the "vertical growth" of Anthropic, but its massive compute power and potential acquisitions (like Cursor) make it a player to watch.

Takeaways

• Wait and See: Currently a laggard in business utility, but its connection to the "X" ecosystem and massive capital/compute resources keep it in the conversation.