Investors have been fixated on the relative performance of Alphabet's AI models when they should instead be cheering the company's cloud potential, an analyst says

MoffettNathanson projects Google Cloud revenue could grow at a 48% compound annual growth rate through 2030.

Shares of Alphabet have fallen 15% off their May highs as the company's artificial-intelligence capabilities come under scrutiny. Yet the fixation on AI leaderboards may be missing the point, according to MoffettNathanson analyst Michael Nathanson.

High-profile departures from the company's Google DeepMind lab, combined with a delayed Gemini 3.5 Pro model, have led many to wonder if Alphabet (GOOGL) (GOOG) is losing its cutting-edge AI capabilities. Nathanson, though, believes that Alphabet is playing a different game than competitors like OpenAI and Anthropic.

That's because Alphabet's large cloud-computing business may hold the key its success in the AI era, rather than whether or not the company has the most high-performance AI model, Nathanson wrote in a Wednesday note.

"We do not dispute the gaps on several benchmarks," Nathanson said of Gemini's performance. "Instead, we disagree with the conclusion that this is necessarily bearish for Alphabet."

Alphabet is instead choosing to funnel resources to its Google Cloud business, which Nathanson believes is a more profitable strategy than "chasing frontier leadership at any cost."

Alphabet is uniquely positioned across the entire AI tech stack, spanning from its custom chips to distribution surfaces like YouTube and Android. Maintaining its full-stack capabilities is a better use of money than making Gemini a frontier model, Nathanson wrote.

On the latest earnings call, Alphabet shared that over half of the company's machine-learning computing power is expected to go to the cloud business in 2026, and Nathanson expects that percentage to only grow from here.

That's not to say that Google will forgo its Gemini ambitions. Alphabet's best strategy is to keep Gemini "good enough to remain at or near the frontier" while investing more into Google Cloud, he added. While the latest Gemini Pro model remains delayed, Google has been releasing more cost-efficient Flash models.

MoffettNathanson projects Google Cloud revenue will surge to $415 billion by 2030, representing a 48% compound annual growth rate.

It's unclear if the AI business model will result in a single provider dominating the market, or if there will be room for multiple winners. But for Alphabet, prioritizing the cloud business appears to be the "best overall probability-weighted opportunity to win," according to Nathanson.

A world with multiple AI winners would be most favorable to Alphabet, as it would result in high demand for its various infrastructure and enterprise offerings.

"Model leadership is increasingly expensive to maintain and...increasingly short-lived," Nathanson said. Meanwhile, Alphabet's cloud capacity can be monetized regardless if the winning model is Gemini or another one. According to MoffettNathanson's projections, Anthropic alone could account for 20% of cloud revenue in 2028.

"The objective should be to maximize returns on AI investment, not simply time spent at the top of the leaderboard," Nathanson said.