AI’s Hard Economics for Good Times: Navigating pricing, costs, and monetization

Orb and Baseten hosted a panel of AI industry experts to tackle the complex challenges of pricing, costs, and monetization in the rapidly evolving AI landscape. The event brought together leaders from various segments of the AI market, including Matan-Paul Shetrit from Writer, our own Kshitij Grover of Orb, Joshua Ma from Airtable, Evan Conrad of San Francisco Compute, and Tuhin Srivastava from Baseten, with Corinne Marie Riley from Greylock moderating the discussion.

The big questions on the table:

• How should we think about costs when it comes to building with AI?
• Once we’re in the race, how do we manage the trade offs between being at the forefront of innovation with spend and performance? And;
• How are we thinking about pricing strategies, especially as companies transition across various stages of maturity?

The panel brought plenty of insight and diverse perspectives, and despite their different approaches, everyone agreed on one thing: AI companies are facing some truly unique challenges and opportunities in today’s market.

Below, you'll find a recap of the event summarizing the takeaways and learnings shared by the panel.

Balancing Innovation and Revenue

One of the central themes of the discussion was the delicate balance between continued innovation and the pressure to generate revenue. As companies transition from pure R&D to commercialization, they face the challenge of justifying AI investments while demonstrating tangible business value.

Josh from Airtable emphasized the importance of aligning AI capabilities with customer-perceived value: "As long as you're tracking closely to how our customers think of the value they're actually getting from the product and the business processes we're making more efficient, I think that will go a long way."

This sentiment was echoed by Kshitij, Co-founder and CTO of Orb, who noted a shift in the AI startup landscape. Unlike traditional SaaS companies, many AI startups are now forced to consider costs and pricing models from day one. This early focus on monetization reflects the increasing maturity of the AI market and the growing pressure to demonstrate viable business models.

You can spend a bunch on AI and you can blow up your costs. But, if that's not actually connected to your go-to-market motion and the value that your customers perceive from your product, those two things diverging too much are going to cause trouble at some point."
_{— Kshitij Grover, Co-founder & CTO at Orb}

The Shifting Goalposts of AI Pricing Models

As the AI market matures, companies are experimenting with various pricing strategies to find the right balance between value delivery and cost recovery. The panelists discussed several emerging trends in AI pricing:

1. Prepaid Credits: Kshitij highlighted the growing popularity of prepaid credit models, which offer flexibility for customers and simplify conversations with procurement teams.
2. ‍Value-Based Pricing: Moving beyond simple usage-based pricing, Matan from Writer stressed the importance of pricing based on the value delivered to customers: "Where you really start deriving strong economics as a company is moving away from just purely charging on a model usage basis, but actually providing outsized value to your customers."‍
3. Hybrid Models: Several panelists suggested that the future of AI pricing might lie in hybrid models that combine fixed and usage-based components, allowing companies to balance predictable revenue with the ability to capture upside from high-value use cases.

Enterprise Sales in an AI Era

Panelists agreed that selling AI solutions to enterprises presents unique challenges and opportunities. Matan emphasized the importance of finding executive sponsors and proving value with initial use cases before expanding within an organization. This "land and expand" strategy is particularly crucial in the AI space, where the technology's potential impact can be vast but may require significant organizational buy-in.

Tuhin offered an interesting perspective on accelerating enterprise sales cycles by offering on-premises deployments, which can bypass lengthy security reviews. This approach highlights the need for AI companies to be flexible in their deployment models to meet the varying needs of enterprise customers.

The Build vs. Buy Dilemma in AI Development

An intriguing point of discussion centered around the decision to build AI capabilities in-house versus leveraging external solutions. Matan advocated for a pragmatic approach:

I would buy anything I can to get to market as quickly as possible. I think people in the Valley especially are too in love with building and forgetting that we need to prove our business."
_{— Matan-Paul Shetrit, Director of Product at Writer}

This perspective underscores a growing recognition in the AI industry that speed to market and proving business value often outweigh the desire for complete control over a technology stack. It also points to the maturing AI ecosystem, where specialized vendors are emerging to provide key capabilities that can be integrated into broader solutions.

Generalized vs. Specialized: the Future of AI Models

The panel concluded with a thought-provoking discussion on the future direction of AI models. While large, generalized language models have dominated headlines, the panelists saw a growing role for specialized, domain-specific models.

Matan shared the Writer philosophy on the topic: "We don't think there's going to be a universal large model that's going to cost a hundred billion dollars. The same way that you don't go to your dentist when you want to fix your car, you're not going to use a generalist model or a med model when you want to deal with financial concepts."

Tuhin suggested a more nuanced approach, proposing that the future might involve a combination of generalized and specialized models: "You're going to use general purpose APIs and providers for maybe anywhere between 50 to 90 percent of your workload, and for those really special use cases that might be the highest value use cases, you're looking to do something a bit more domain specific."

This perspective highlights the ongoing evolution of AI technology and suggests that successful AI companies will need to be strategic in how they leverage different types of models to deliver maximum value to their customers.

Companies in the AI industry face complex decisions around pricing, cost management, and technology strategy. The speakers all agreed on the importance of focusing on customer value, remaining flexible in approach, and continuously innovating not just in technology, but in business models as well. By navigating these challenges successfully, AI companies can position themselves to capture the enormous potential of this transformative technology while building sustainable, profitable businesses.

Want to see the full event? Check out the video below to hear the discussion in full.