How to Invest in AI agents

What is an AI agent and how does it work?

An AI agent is software that uses a large language model as a reasoning engine to plan and carry out multi-step tasks, rather than answering a single prompt. Where a chatbot returns text, an AI agent breaks a goal into steps, calls tools and APIs, checks its own results, and loops until the task is done. That tool-calling loop is the core mechanism behind the AI agents theme: it is what turns a model into something that can file a ticket, reconcile a spreadsheet, or draft and route a contract.

The companies in the AI agents theme sit at different layers of that loop. Model providers reached through MSFT (its OpenAI partnership) and GOOGL (Gemini) supply the reasoning engine. Enterprise platforms like CRM, NOW, and PLTR provide the workflows, data, and permissions an agent needs to act safely inside a business. NVDA-class compute, reached here through the platform owners, runs the inference underneath. The AI agents theme groups the firms positioned to monetize this shift from answering questions to completing work.

Why are AI agents emerging as an investment theme now?

AI agents are surfacing as a distinct theme now because models have become reliable enough at tool use and long-context reasoning to chain many steps without losing the plot. Earlier models could draft text but stumbled when asked to take a sequence of real actions; the current generation, reached through MSFT and GOOGL, is markedly better at calling tools and recovering from errors, which is the threshold the AI agents theme depends on.

The second driver is distribution. CRM, NOW, and MSFT already sit inside the systems where work happens, so they can attach agents to existing seats and data without a customer rebuilding anything. PLTR pushes agents into operational decisions at large enterprises and governments. This is why the AI agents theme is led by incumbents with workflow ownership rather than pure startups: the agent is only as useful as its access to a company's tools and data.

How will AI agents make money?

Monetization in the AI agents theme is forming around two models. The first is per-seat or per-agent subscription pricing, where MSFT (Copilot), CRM (Agentforce), and NOW attach an agent to an existing license and charge for the added autonomy. The second is consumption pricing, where customers pay for the tokens and tool calls an agent consumes, which flows back to the model and compute owners reached through GOOGL and MSFT.

PLTR sits slightly apart, selling agentic capability bundled into high-value operational deployments rather than commodity seats. As of 2026 these models are still being proven, so the AI agents theme carries genuine uncertainty about which pricing structure wins and how durable the early revenue is. That early-stage character is exactly why the theme reads as a satellite tilt rather than a settled, core allocation.

What gets a stock into the AI agents theme?

Revenue exposure to autonomous, tool-calling AI agents: frontier model providers, enterprise platforms shipping agentic products into existing workflows, operational AI deployment specialists, and the compute the agents run on.

What stocks are in the AI agents theme?

Every public name that fits the AI agents thesis, with the rationale for inclusion. Click any ticker for the full stock guide. The basket above starts equal-weighted; you set your own target weights inside Walnut.

How to invest in AI agents

There are a few ways to get exposure to the AI agents theme, and Walnut is not an investment adviser, so this is descriptive rather than a recommendation. The most concentrated path is buying individual stocks that fit the thesis directly: enterprise platform owners like CRM and NOW, the model-and-distribution incumbents MSFT and GOOGL, and operational deployment specialist PLTR. Picking single names lets you weight the exact layer of the agent stack you find most compelling, at the cost of more research and more concentration in a still-forming theme. The passive route is broad technology ETFs: QQQ holds the large-cap agent incumbents, while VGT and XLK give wider software and hardware exposure. The tradeoff is dilution, since none of these is a pure-play AI agents vehicle and the agent exposure is a small slice of each fund.

The third approach is building a dedicated AI agents basket in Walnut. You describe the thesis to Walnut's AI assistant, for instance AI agents spanning model providers, enterprise workflow platforms, and the compute beneath them, and the assistant proposes constituents and starting weights drawn from names like MSFT, GOOGL, CRM, NOW, and PLTR, with the rationale for each. You review every constituent and weight, adjust anything you want, and fund the basket through your own connected broker. You approve every order before it is placed; Walnut never trades on your behalf. The AI agents basket then tracks as a single performance line you can compare against QQQ or VGT.

Which ETFs cover AI agents?

If you want the theme as a single ticker rather than as a basket, these are the ETFs people most commonly use. Each has trade-offs (concentration, expense ratio, sector overlap) covered in the individual ETF guides.

The bottom line on AI agents

AI agents extend the AI software story from chat into autonomous, multi-step work, and the early beneficiaries are the platforms that already own enterprise workflows: MSFT, GOOGL, CRM, NOW, and PLTR. Because the pure-play agent layer is still nascent and overlaps heavily with broad software, the theme fits a portfolio as a satellite tilt. A focused basket expresses it more cleanly than QQQ or VGT, where the agent exposure is buried inside hundreds of unrelated names.

FAQ