Robert F. Smith believes the next phase of the artificial intelligence boom will not be defined by data centers, chips or hyperscalers, but by private enterprise software companies quietly embedding AI into real business workflows. Speaking amid growing concern that parts of the AI infrastructure trade may be overheating, Smith argues that productivity-driven application providers will ultimately capture the largest share of long-term value. From the perspective of NewsTrackerToday, this marks a decisive shift in how the AI cycle is beginning to be priced.
Much of the public market conversation remains dominated by companies such as Nvidia, Microsoft, Alphabet and Meta Platforms, alongside frontier model developers like OpenAI. Smith does not dismiss their importance, but he frames them as infrastructure builders rather than ultimate value owners. In his view, the economic rent in AI follows a familiar pattern: hardware enables capability, platforms distribute it, and applications monetize it once adoption becomes operational.
Vista Equity Partners, which manages more than $100 billion in assets and controls over 90 enterprise software companies, is positioning itself squarely in that application layer. Smith notes that roughly 97% of enterprise software companies remain private, making them largely invisible in public-market narratives about AI winners. NewsTrackerToday sees this as a structural blind spot for investors who equate AI progress with listed mega-cap performance.
Central to Vista’s strategy is what Smith describes as an internal “agent factory” – a repeatable framework for deploying agentic AI across its portfolio. According to Smith, around 30 Vista-backed companies are already generating revenue from agent-based AI systems, with another 30 to 40 expected to follow in the coming months. These agents are designed not to answer questions, but to complete tasks end-to-end: documenting workflows, assessing regulatory exposure, managing compliance and executing routine but high-friction business processes.
Liam Anderson, financial markets analyst at NewsTrackerToday, views this as the most investable part of the thesis. “The infrastructure layer prices expectation, but the application layer prices evidence,” he says. “If agentic AI consistently reduces labor hours, error rates and cycle times, that’s where sustainable margins show up – regardless of whether infrastructure spending eventually cools.”
Smith points to portfolio examples such as SimplePractice, where AI agents automate clinical documentation for mental health professionals, and Resilinc, which uses agents to model tariff exposure and regulatory risk. In both cases, AI is embedded into systems of record rather than bolted on as a standalone feature. From NewsTrackerToday’s perspective, this distinction matters: tools that own workflows are far harder to displace than generic AI interfaces.
The approach also challenges the long-standing idea that AI would “consume” traditional software, a notion famously raised years ago by Nvidia’s leadership. Smith argues instead that AI expands the value of enterprise software by allowing it to orchestrate services rather than merely provide interfaces. Productivity gains of 30% to 50% in coding and task execution are already being observed across Vista’s portfolio, according to Smith, with low-cost AI inference producing outsized economic returns.
The labor implications are unavoidable. Smith acknowledges that some roles will disappear, but he frames the broader impact as reallocation rather than collapse. Workers who fail to adapt will be displaced, he argues, while those who learn to operate alongside AI agents will see their leverage increase. NewsTrackerToday notes that this framing aligns with early enterprise adoption patterns, where AI reduces routine effort while increasing demand for oversight, judgment and system design.
Looking ahead, NewsTrackerToday expects 2026 to become a test year for this thesis. If infrastructure spending begins to normalize while application-level profits continue to scale, investor attention is likely to migrate rapidly toward private enterprise software and away from pure compute narratives. The key differentiator will not be model performance, but proof of recurring productivity and margin expansion.
The implication is straightforward. AI’s next winners are unlikely to be the loudest or the most capital-intensive. They will be the companies that quietly turn autonomy into measurable business outcomes. As News Tracker Today sees it, Robert F. Smith is not predicting the end of the infrastructure boom – he is signaling where the durable economics of AI are likely to settle once the cycle matures.
