Palantir Technologies (PLTR 6.25%) was the best-performing stock in the S&P 500 last year. The company began to see a lot of momentum with its artificial intelligence (AI) platform, as its focus on the workflow and application layers of AI led many companies to test its solution. It continues to have a big opportunity as it starts transitioning customers from proof of concept into production.
That said, the strong run in Palantir's stock has left it with an astronomical valuation, trading at a forward price-to-sales ratio (P/S) of 40 times fiscal 2025 estimates for a company that just grew its revenue by 30% last quarter. That's more than double peak software-as-a-service (SaaS) multiples from a few years ago when SaaS stocks were growing in the mid-30% range. Meanwhile, Palantir executives, including its CEO, chairman, and chief technology officer, among others, have been aggressively selling shares the past few months.
Image source: Getty Images.
Let's look at two other stocks benefiting from AI trading at much cheaper valuations that investors can consider.
Nvidia
Nvidia (NVDA 4.45%) has been one of the biggest AI beneficiaries, as the graphic processing units (GPUs) it designs have become the backbone of AI infrastructure. As a result, the company's revenue skyrocketed, including it producing 94% revenue growth last quarter.
While its stock soared over the past few years, it nonetheless trades at attractive valuation with a forward price-to-earnings (P/E) ratio of under 31 based on 2025 analyst estimates, and a price/earnings-to-growth (PEG) ratio of approximately 0.96. A PEG ratio less than 1 is generally viewed as undervalued, but growth stocks will often have PEG ratios well above 1.
NVDA PE Ratio (Forward 1y) data by YCharts
Nvidia still has a big opportunity in front of it. As tech giants and AI start-ups race to build ever-better AI models, they need exponentially more computing power, and thus GPUs, to train these models on. While Meta Platforms' Llama 3 model was trained on 16,000 GPUs and xAI's Grok 3 was trained using 20,000 GPUs, Meta's Llama 4 model is being trained on 160,000 GPUs and xAI's Grok 4 is being trained on 200,000 GPUs. Meanwhile, there is talk of future AI models being trained using clusters of 1 million GPUs in the not too distant future.
As the GPU leader, Nvidia is poised to continue to greatly benefit from this AI infrastructure buildout. Its CUDA software platform helped it establish a wide moat in the space, as long ago it was the first GPU company to introduce software to allow its chips to be programmed for tasks other than graphics rendering. As such, CUDA became the de facto platform on which developers learned to program these chips. With the introduction of a number of AI-specific microlibraries and developers tools, CUDA continues to be a huge differentiator for the company.
Salesforce
On the AI software side, Salesforce (CRM 0.68%) stock is an attractively priced alternative to Palantir, trading at 7.7 times next year's analyst estimates.
CRM PS Ratio (Forward 1y) data by YCharts
While its valuation is much lower, the company has a big potential AI opportunity in front of it with its Agentforce solution. Salesforce is aiming to become the leader in agentic AI, which is widely considered the next step beyond generative AI. With agentic AI, AI agents will autonomously work within parameters to perform complex tasks on their own with little human oversight needed. While generative AI will tell you the best way to do something, agentic AI will just go out and do it itself.
With its Agentforce solution, Salesforce says its AI agents can be used for a number of tasks across industries. For example, in retail its autonomous AI agents can learn from shoppers' behavior and preferences and act as a digital concierge to converse with them in a natural way. Meanwhile, it can be used for customer service in various industries to address billing complaints by reviewing past bills, identifying problems, and validating disputes.
Agentforce is a usage-based product that costs $2 per conversation. On its early December earnings call, Salesforce said it had already closed 200 Agentforce deals since its launch in October and that it had thousands of potential deals in its pipeline. It forecast deploying 1 billion Agentforce AI agents by the end of fiscal 2026 (ending January 2026).
Meanwhile, the company is innovating quickly, releasing Agentforce 2.0 following its fiscal third-quarter earnings report. New features with Agentforce 2.0 include a no-code platform to let users build their own tools, the ability to work outside of Salesforce's suite of products, and integration into its Slack product.
While Salesforce has not been growing as quickly as Palantir, it nonetheless looks to be at the start of a huge AI opportunity and trades at a fraction of the valuation.
