Smith opened by framing enterprise software through the lens of his engineering background at Bell Laboratories, where full hardware-software-firmware solutions first demonstrated how technology removes inefficiencies. He traced the shift from episodic, manual process control to continuous digital observation and intervention, with every gap in the control cycle representing waste that software eliminates.
As silicon costs collapsed under Moore's Law, that productivity logic spread from factories into offices, extracting waste across the global economy. Over time, the tech stack split between hardware and software, with software ultimately capturing the bulk of long-term economic value. Smith called enterprise software the most productive tool introduced to the business economy in the last 50 years.
Software is now entering its third paradigm: from product, to software-as-a-service, to software-as-a-worker. Agentic systems operate continuously inside enterprise workflows, working around the clock, compensated in compute tokens rather than benefits, and capable of absorbing more context than human workers in many tasks.
Not every software company will earn what Smith calls the "right to thrive." Businesses built only on publicly available content can be replicated by efficient AI search and lack defensibility. The winners will be those with sovereignty and dominion over proprietary workflows and data sets. He noted that less than 1% of enterprise data sits in public environments where large models have been trained, giving incumbents with decades of context, customer trust, and industry reach a structural advantage.
Echoing Satya Nadella's argument from Davos, Smith stressed that enterprises should bring models to their data rather than export their data to external models, to avoid leaking intellectual property and value into someone else's system. The vault stays inside the enterprise; the inference comes to it.
On the investment opportunity, Smith pointed out that 97% of software companies are private, giving Vista a vast universe beyond public markets. Vista has outperformed public markets across its 26-year history, and now manages over $110 billion across private equity and credit, with both arms informing each other on which companies have the right to thrive under AI.
Vista has built its own "agentic factory," partnering with hyperscalers to develop industry-specific agents and deploy them directly into customers' environments. Smith expects gradual adoption followed by sharp industry-wide spikes, shifting markets from "winner-take-most" to "winner-take-all" because agent capacity is bounded only by available compute, not by human limits.
Leadership style will determine adoption speed. Smith argued that product-oriented CEOs will adopt agentic AI faster than operations-focused leaders, and the more productive AI becomes, the more it gets used, creating a compounding cycle of capability and demand.
The real rate-limiting constraint, in Smith's view, is the cost of inference and the compute cycles required to run agents at scale. Current infrastructure is built primarily for training rather than inference, making it energy-inefficient for ongoing agent operations. Until purpose-built inference infrastructure scales, energy consumption will cap how many agents enterprises can run continuously.
For GCC investors thinking across generations, Smith offered three principles: engage strong, curious, globally-minded advisors who can interpret a fast-moving landscape; be willing to take measured risk in less-understood areas where pricing arbitrage still exists before the upside becomes widely known; and look for institutions that know how to drive transformation and value conversion using AI, rather than buying assets at peak valuations.
Watch the full webinar above for more details.
