The evolving preference of AI agents for foundational models is redefining who wins in the SaaS and AI landscape. While OpenAI shows signs of revival, Anthropic races for compute support, and private equity giant Thoma Bravo faces a $5.1 billion equity wipeout on Medallia.
- AI agents now dictate model and workflow preferences, shifting SaaS industry dynamics.
- OpenAI rebounds with models favored by autonomous agents despite earlier missteps.
- Thoma Bravo wipes out $5.1 billion equity on Medallia amid changing PE market conditions.
What happened
The AI and SaaS markets are experiencing a fundamental shift as autonomous AI agents are increasingly driving the selection of foundational models, vendors, and workflows. This transition marks a departure from human-centric preferences and introduces a new competitive landscape where models that best serve these AI agents gain market share. OpenAI, which reported weaker-than-expected Q4 results, is showing signs of a comeback as its models become preferred by these agents over Anthropic’s Claude, which had an early lead in human popularity due to its coding capabilities.
In parallel, Anthropic secured a massive $45 billion investment from Google and Amazon to support its compute demands, highlighting the intense resource race in the AI model space. This comes as Nvidia and Google hit significant market valuations of $5 trillion and $4 trillion respectively. Additionally, private equity firm Thoma Bravo suffered a staggering $5.1 billion equity wipeout on its Medallia investment, reflecting broader market recalibrations and challenging legacy PE approaches in the AI-enabled B2B arena.
Why it matters
The shift to agent-selected models redefines competitive advantages and the value proposition of SaaS companies. Human preferences that guided previous AI-era investments are losing relevance as AI agents become the true end users. This transformation affects foundational model wars, software vendor viability, and ultimately, B2B company terminal value. Companies that cannot attract AI agent usage risk falling into lower valuation categories as AI-first models reshape market expectations and customer workflows.
This trend also alters investment and IPO dynamics in the SaaS sector. The much-anticipated ‘SaaS pixie dust’ premium—valuations soaring to 30x revenue—has evaporated in 2025. Investors now require clear evidence of value creation alongside scale and profitability. Companies like Canva face scrutiny over whether their products genuinely attract AI agent users. The fallout for private equity is profound, as demonstrated by Thoma Bravo’s Medallia loss, signaling a need for strategic rethink of B2B SaaS buyout strategies amid an AI-first economy.
What to watch next
Watch closely how AI agent adoption influences future model leadership and vendor success. OpenAI’s regained favor among AI agents could indicate a sustainable advantage if it continues to deliver superior workflows and integration capabilities. Meanwhile, Anthropic’s heavy reliance on external capital to fuel compute capacity raises questions about its long-term competitiveness if AI agents do not align with its models. The evolving preferences of these agents will be a critical bellwether for B2B SaaS vendors seeking to maintain or grow their market relevance.
Investors and operators should monitor how the new AI-centric framework impacts public market valuations and PE strategies. SaaS companies that fail to design for and attract AI agents risk being relegated to lower-value categories. IPO-stage companies must demonstrate strong fundamentals beyond growth, emphasizing profitability and AI adoption. The broader industry must adapt to a landscape where agent-centric automation is the norm, recalibrating everything from product development to investment thesis accordingly.