Investors should think of artificial intelligence (AI) in three phases. The first phase involves high-performance semiconductors and networking gear required to build supercomputing infrastructure. Nvidia has dominated the first phase, but chipmakers like Broadcom and networking specialists like Arista Networks have also benefited.

The second phase involves cloud providers that offer infrastructure and platform services needed to build conversational copilots, autonomous agents, and other AI applications. Public clouds like Microsoft, Amazon, and Alphabet have been winners in the second phase, as have platform providers like Palantir.

The third phase involves software vendors that combine products from the first two phases with their proprietary data and technical know-how to build AI applications. UBS analysts believe this phase ultimately offers the largest monetization opportunity. The phases are not mutually exclusive, but they do build on one another.

That we are entering the third phase does not mean the first two phases have ended, but the third phase would not be possible without the underlying hardware and cloud services. One company Wall Street likes in the AI software space is Salesforce (CRM -0.96%). The stock has a consensus rating of buy, and a median target of $415 per share. That implies 24% upside from the current share price of $336.

Here's what investors should know.

Salesforce recently introduced new agentic AI software

Salesforce has dominated the customer relationship management (CRM) market for over a decade. It had nearly 22% revenue share last year, more than the next four rivals combined, according to the International Data Corp. (IDC). The company also ranked first in several individual CRM software verticals, including sales, customer service, and marketing.

Salesforce earlier this year announced Agentforce, a platform that supplies digital labor powered by artificial intelligence (AI). The utility of Agentforce extends beyond that of conversation copilots by "using advanced reasoning abilities to make decisions and take action, like resolving customer cases, qualifying sales leads, and optimizing marketing campaigns," according to the press release.

To date, Salesforce has launched two agentic AI products: Agentforce Service Agent and Agent Builder. The former is a customer-facing conversational interface that can automate customer service workflows; the latter lets business build custom AI agents. However, the company is still innovating. Salesforce recently announced new features and use cases, collectively called Agentforce 2.0, most of which will launch by February 2025.

CEO Marc Benioff believes Agentforce 2.0 positions Salesforce firmly ahead of competitors like Microsoft in the race to automate labor and improve worker productivity. "The demand for Agentforce has been amazing," he commented in the press release. "No other company comes close to offering this complete AI solution for enterprises."

Salesforce saw early momentum with Agentforce in the third quarter

Salesforce reported solid financial results in the third quarter of fiscal 2025, which ended in October 2024, despite missing bottom-line estimates. Revenue rose 8% to $9.4 billion on particularly strong growth in sales and customer service software, and non-GAAP earnings increased 14% to $2.41 per diluted share.

Brian Millham, Salesforce president and chief operating officer, mentioned strong demand for Data Cloud and Agentforce on the third-quarter earnings call. Data Cloud is the foundational layer that aggregates information from other clouds (i.e., sales, service, and marketing). That data then informs decisions made by autonomous agents.

Currently, 25% of Fortune 100 companies are Data Cloud customers, which paves the way for strong adoption of Agentforce in future quarters. Indeed, the company closed over 200 Agentforce deals in the third quarter, despite the platform being released days before the quarter ended. And Benioff told analysts, "The pipeline is in the thousands."

Shareholders should be encouraged by that news. Spending on AI agents is forecast to grow at 45% annually through 2030, according to Grand View Research.

Irridescent text bubbles that read "AI" shown on a digital screen.

Image source: Getty Images.

Salesforce stock is expensive, but investors should still consider buying

Wall Street estimates Salesforce's adjusted earnings will increase at 12% annually through fiscal 2026, which ends in January 2026. That consensus estimate makes the current valuation of 35 times adjusted earnings look fairly expensive. Having said that, analysts may upwardly revise earnings estimates if Agentforce sees particularly strong adoption.

To that end, investors comfortable with volatility can buy a small position today. However, I also think better buying opportunities will present themselves in the future. So, prospective investors and current shareholders interested in building larger positions should be ready to buy the stock on the dip. Personally, I would be more interested in Salesforce if the share price fell by 15% to 20%.