Best Ideas 2026

AI: Achieving Scale

Technologic sphere.

Artificial intelligence is the most far-reaching technological development since the creation of the internet. We believe AI represents a generational paradigm shift in how consumers and enterprises interact with and use computing services. For enterprises, AI offers enhanced efficiency, superior execution, strategic differentiation, and deeper insights. For consumers, AI provides instantaneous access to information, personalized content experiences, and advanced problem-solving capabilities. 

AI’s potential has attracted investors, who have bid up AI-related stocks and pledged hundreds of billions of dollars to support AI capabilities. This has resulted in unprecedented market concentration and some skepticism, but we are not among the doubters. Our observations over the past year have confirmed our view that AI continues to develop effectively and rapidly, with enormous implications for the AI ecosystem and global economy. In fact, the extraordinary investments in AI infrastructure are already yielding results. The foundation for agentic AI applications is now in place, allowing consumers, creators, and enterprises to benefit from rapidly improving AI capabilities:

  • Consumers are increasingly interacting with AI within various products across mainstream applications—everything from search to Instagram. Users have been able to ask increasingly sophisticated questions and receive more detailed, appropriate, and helpful responses.
  • Multimodality models, which include images and video, offer advanced tools for creators. Advertisers, for example, can take one piece of media and create multiple variations that can be tailored to audiences.
  • For enterprises, agents are now capable of doing longer-form tasks—in some cases doubling their capacity from 12-24 months ago—and workflows are being automated into minute-long processes that once took hours, if not weeks. This applies to a range of applications—from creative solutions to mundane, repetitive tasks—and frees employees to focus on higher-quality projects.

The growing sophistication of AI applications, and their growing use across the economy, has shifted investor focus increasingly to monetization. Monetization has already occurred for the companies that build the components of the AI infrastructure (e.g., Nvidia) and now, with the very large and fast deployment of AI infrastructure over the past two years, hyperscalers are well positioned. They will rent out increased compute capacity to customers (e.g., software companies, enterprises building custom solutions, consumers), where demand remains strong. We believe this will result in higher revenue for the hyperscalers and represents a transition in the deployment and monetization of AI.

Most importantly, monetization is occurring for the companies that develop AI applications and services—more than a dozen AI companies have gone from virtually zero to a billion dollars in annual recurring revenue over the last 18 months. We believe these companies have the potential to be more profitable over the long term, but currently their operating margins are being lowered by the high costs of compute. As costs for AI models get cheaper, in our view, and these model companies are able to deliver more intelligence more cheaply, profit margins should improve.

 

More Efficiency, More Opportunity

Given the capabilities of AI agents today, one significant challenge for companies is getting their data organized so it can be used effectively by AI tools. We believe many firms will address this issue in 2026. This will be a meaningful opportunity for software companies that provide AI solutions and for the companies themselves, which should see significant efficiencies and opportunities from AI analysis.

We note “bubble-like” behavior in parts of the market, but the demand across the AI ecosystem is healthy and will continue at least in the short term. In prior investment cycles, such as the internet around 2000, a large amount of infrastructure was built but not used. In contrast, almost all of the GPU-related infrastructure is being utilized, resulting in well-publicized demand surges for AI components, expertise, and energy. 

However, we are monitoring a few trends because of their potential impact on the AI ecosystem:

  • AI models could become dramatically more efficient, cheaper to run, and require less compute to handle inbound requests or tasks. A step-change improvement could lower demand for, among other things, AI infrastructure and energy. 
  • Open-source models have become very effective, especially for certain tasks. If open-source models, which are free, offer capabilities that rival those from AI model providers, it could drive a revaluation of AI investments and prospects. 
  • AI continues to require enormous capital expenditures, and one of the major barriers to AI is the compute resources needed. Looking ahead, we believe the growth of capital expenditure is sustainable. The leading frontier model providers are very profitable and can, in our view, fund capex from operating cash flows. In addition, we believe the capex investments could result in major drivers of revenue growth which, in turn, could fund the investments.

Compared to prior waves of technology development, AI is still in a very early stage. However, AI is accelerating at a faster rate in historical terms. ChatGPT grew to 800 million weekly active users in two-and-a-half years; it took the internet about 12 years to achieve the same scale. In this environment, we believe it is critical to maintain perspective. Secular growth themes can stoke market enthusiasm and boost valuations but, in our experience, markets ultimately separate winners from losers. We have been applying our growth investing approach for more than five decades, and we believe that our focus on fundamentals, deep resources, and experience across multiple cycles of economic transformation leave our clients well positioned to benefit from the AI-driven revolution.

Related Strategies

Fundamental Equity

Jennison's investing approach is rooted in fundamental research and security selection.

Learn More