Artificial intelligence is moving rapidly from innovation theme to capital cycle, with investment opportunities broadening well beyond model developers and semiconductor leaders. The first leg of the trade was defined by the buildout phase—foundational models, accelerated compute, and the infrastructure required to support scale. We are now entering the next phase, where value creation shifts toward inference, agentic software, and sector-specific deployment. In this environment, the likely winners will be companies that can pair proprietary data with disciplined execution to drive productivity gains, margin expansion, and durable competitive advantage.

The first wave of AI investment was concentrated in the foundational layer of the stack. Private market enthusiasm centered on model developers such as OpenAI, Anthropic, and Perplexity, while public market leadership was driven by enabling hardware, most notably Nvidia. That was followed by the infrastructure buildout required to support adoption at scale, including data center capacity, power demand, and the broader ecosystem needed to sustain higher compute intensity.

The market is now moving beyond buildout and into deployment. AI has matured to the point where inference—the stage at which trained models generate real-world outputs—becomes the key commercial layer. That shift enables the rise of agentic systems capable of executing tasks, automating workflows, and enhancing decision-making across industries. For investors, the question is increasingly straightforward: which companies can deploy inference and agents in ways that materially improve their competitive positioning?

The implications are broad. In customer service, AI has the potential to fundamentally reshape call center economics and service quality. In consumer markets, intelligent agents could evolve into personalized digital concierges. Across the enterprise, agentic systems should reduce friction in business processes, accelerate automation, and improve operating efficiency. In transportation, they are likely to deepen the functionality of autonomous driving and robotaxi platforms. In healthcare and life sciences, they may shorten development cycles and improve the economics of drug discovery. The addressable opportunity is expanding rapidly, and the number of viable use cases continues to grow.

At the same time, not all AI deployment will create equal value. One of the clearest differentiators will be data quality. As the principle of “garbage in, garbage out” applies with even greater force in the AI era, companies with proprietary, high-quality, well-structured data should be better positioned to generate superior outcomes. In practice, that means sustainable advantage is likely to accrue not only to firms with strong technical implementation, but also to those with differentiated data assets and the organizational discipline to deploy them effectively.

We believe 2026 is likely to be the year in which these themes begin to move from narrative to measurable market impact. As pilots become production systems and enterprise use cases scale, investors should gain better visibility into where AI is truly driving revenue acceleration, productivity improvement, and margin leverage. That transition should help separate early enthusiasm from durable value creation.

Looking further ahead, late 2026 and 2027 could mark the beginning of a new company formation cycle built around AI-native architectures. As in prior technology transitions, a new cohort of firms is likely to emerge with operating models designed for the capabilities of the new platform, while slower incumbents risk disruption. Over time, that dynamic should extend the opportunity beyond enablers and adopters to a new generation of AI-native businesses, with the IPO market eventually providing a clearer window into the next class of category leaders.