The AI Trillion-Dollar Question: Which AI Agent Company Will Break the SaaS Ceiling

AI Summary: Bret Taylor's core message was that the AI revolution's biggest disruption isn't just technological, but in business models, offering startups a significant advantage over incumbents burdened by legacy structures. He highlighted outcomes-based pricing as a transformative third evolution in software business models, following perpetual licenses and SaaS, arguing that applied AI companies focused on solving specific, high-value vertical problems and selling "outcomes" rather than just tools are poised to create trillion-dollar opportunities and reshape market share, much like the internet era.

May 13 2025 17:23
At AI Ascent, Bret Taylor – former CTO of Facebook, co-founder of multiple companies, and current co-founder of Sierra – delivered a prescient message: the AI revolution isn't just changing technology; it's fundamentally reshaping business models in ways that could topple tech giants and create unprecedented opportunities for startups.

"Most people in this room are not encumbered by a business model," Taylor told the attendees, "Which is a funny way of thinking about it. But if you look at the history of software, it's actually much harder to change your business model than to close a technology gap in your product."

Taylor's company Sierra helps businesses build customer-facing AI agents. Its pricing approach is surprisingly straightforward for what might seem like complex technology: if the AI resolves a customer issue autonomously, Sierra gets paid a pre-negotiated rate. If the issue needs escalation to a human, Sierra doesn't charge.

This "outcomes-based pricing" represents a third major evolution in software business models, following the shift from perpetual licenses to subscription-based SaaS. According to Taylor, it's a natural progression for AI companies that "complete a job" rather than just provide tools – and it may create a massive advantage for startups over incumbents that remain stuck in traditional subscription models.



History shows us that transitions between business models can be brutal for established companies. Taylor points out that while Salesforce beat Siebel Systems and ServiceNow outperformed legacy ITSM vendors, the real challenge wasn't technical.

"Closing a technology gap in your product is hard but not impossible. Changing your business model is really hard," Taylor explained. "There's a graveyard of CEOs who've been fired because they couldn't make that transition – in part because public company investors are horribly impatient with these things."

He credits Microsoft's Satya Nadella for successfully navigating the shift from Windows to Azure and Adobe's Shantanu Narayen for transitioning to subscription revenue. But these are exceptions that prove the rule: most incumbents fail to make such transitions.

For AI startups competing against established players, this represents a structural advantage that goes beyond having newer technology. Legacy companies may be capable of building competitive AI features, but shifting from charging per seat to charging per outcome requires reimagining their entire business – something few public companies can accomplish without severely disrupting quarterly results.


Taylor divides the emerging AI landscape into three distinct markets with very different dynamics:

• Foundation Models: Capital-intensive businesses like OpenAI that will see significant consolidation, resulting in "a handful of players, relatively low margin, but very high scale" that collect "taxes from everyone in the AI ecosystem" – similar to today's cloud infrastructure providers.
• Tools: The "proverbial pickaxes in the gold rush" – companies making development tools, including existing players like Databricks and Snowflake, plus new entrants. This market faces threats from foundation model companies expanding downstream.
• Applied AI: Companies building vertical-specific AI agents for particular industries and use cases. Taylor believes this is where the trillion-dollar opportunities exist – and where most entrepreneurs should focus.

Why is applied AI so promising? According to Taylor, it changes the fundamental value proposition of software. Traditional enterprise software sells productivity enhancements – making existing processes more efficient. AI agents sell outcomes – actually completing valuable jobs that previously required expensive human labor.

"You're going from selling productivity enhancement to selling outcomes, and outcomes are valuable. Some outcomes are extremely valuable," Taylor explained. "Selling something valuable for an order of magnitude less money than the current cost is really easy."


When asked about horizontal versus vertical approaches to AI, Taylor expressed skepticism about generic AI platforms. "I really believe in verticals in AI," he said, pointing out the vast differences between agents built for telecommunications companies versus commercial banks, insurance providers, or healthcare organizations.

"Every single one of those applications is actually quite specialized," Taylor noted. "The companies that can actually provide value quickly around these core workflows of these businesses have a leg up."

This specialization is particularly important in today's market, where companies are being bombarded with AI pitches that "all sound the same." Taylor compared the current landscape to the early internet era, where many companies were competing with similar-sounding value propositions (remember Altavista, AllTheWeb, and Inktomi competing with Google?).

His advice to entrepreneurs: "Show up as a partner to them and help them solve their acute business problems... do deep research on them. And by that I mean like the actual feature – deep research in ChatGPT... show up educated in these conversations."



For established companies like ADT (American District Telegraph, as Taylor noted) or SiriusXM, AI represents both an existential threat and a transformative opportunity. Taylor believes AI can "really change the cost structure of their business" – particularly for organizations with massive contact centers employing thousands of people.

"These shovel-ready applications of AI, even if indirect, can drive such structural changes to the unit economics of so many companies' businesses that I think it will change market share," Taylor predicted. He compares the current moment to the birth of the internet, where some traditional retailers like Walmart thrived by embracing new technology while others like Blockbuster failed.

Taylor sees an opportunity for established companies willing to embrace AI to substantially reallocate resources:

If you can take all of that opex and put it back into your business, you can lower prices, you can invest in growth.

Perhaps Taylor's boldest prediction concerns the potential scale of companies building industry-specific AI agents. He notes that despite their success, pure SaaS companies like ServiceNow, Salesforce, SAP and Oracle have market caps in the $200-300 billion range – well below the trillion-dollar valuations of infrastructure giants like Microsoft and Amazon.

The key difference, in Taylor's view, is that AI agents aren't just productivity tools – they're fully autonomous workers completing valuable tasks. This fundamental shift in value proposition could allow vertical AI agent companies to break through the ceiling that has limited even the most successful SaaS businesses.


While cost reduction might seem like the obvious selling point for AI agents that replace human labor, Taylor believes revenue growth ultimately proves more compelling for most executives. "If you talk to any CEO, they generally care more about growth than they do cost savings," he noted.

Taylor predicts that most "sophisticated companies will recoup those soft cost savings and reinvest in growth" rather than simply pocketing the difference. Moreover, he warns that cost-saving appeals will eventually lose potency: "Right now most of your AI startups are being compared to labor costs. Ten years from now when it's all AI agents, you're going to be compared to other AI agents."

This perspective reinforces his emphasis on vertical specialization – understanding not just generic cost structures, but the specific business challenges and growth opportunities in particular industries.


Perhaps the most personal insight Taylor shared came from his early days as CTO of Facebook at age 29. "I was sort of trying to do everything myself," he recalled, until COO Sheryl Sandberg delivered a wake-up call about her need to hold his team to higher standards and stop trying to do everything personally. This challenge faces many technical founders as their companies scale.

At every stage of your company, at the beginning, product is all that matters... As you scale, what you need to be great at changes. Having the self-awareness and self-reflection to actually change what you spend your time on is actually one of the greatest challenges.

For entrepreneurs building AI companies, this flexibility may prove particularly vital. As Taylor's insights suggest, success will require not just technical innovation but business model creativity, vertical industry expertise, and adaptability beyond one's core strengths.

While the foundation model companies may capture headlines, Taylor believes the real revolution – and the trillion-dollar opportunities – will belong to those who apply AI to solve specific, high-value problems for particular industries. And in this new landscape, startups unencumbered by legacy business models may have the decisive advantage.