AI turned Google Cloud from also-ran into Alphabet’s growth driver

Audio By Carbonatix

Enlarge

By Kenrick Cai

NEW YORK, NY (Reuters) -Once a money-losing backwater, Google Cloud has become one of Alphabet’s fastest-growing businesses, powered by massive bets on AI and years of costly investment in datacenters, custom chips, and networking gear.

Alphabet’s cloud revenue topped $15 billion in the third quarter, a 34% increase reflecting strong demand for AI infrastructure and services, including Google’s own Gemini model, the company announced Wednesday.

Google Cloud is now challenging YouTube as Alphabet’s No. 2 cash generator behind its search ads business.

“Google Cloud is one of the most important priorities for Alphabet as a whole and I expect it to play an even more central role as the company moves forward,” Alphabet CEO Sundar Pichai told Reuters in an interview earlier in October.

Much of the cloud unit’s growth can be attributed to the business bets and diplomatic maneuvering of its head Thomas Kurian, who joined Google from Oracle in 2018 and grew the unit’s market share from 7% then to 13% in 2025, according to Synergy Research Group, which tracks the cloud industry.

When Pichai replaced Google co-founder Larry Page as CEO in 2019, he identified Google Cloud and YouTube as his two big bets to move Alphabet beyond its core business of search advertising. 

Since then, YouTube has largely delivered, becoming the world’s largest video platform with 1 billion hours watched per day. By contrast, Google Cloud lost billions between 2018 and 2022 due to heavy investments in servers, datacenters and chips until it turned its first profit in 2023.

Now, with generative AI, Alphabet sees a chance to close the gap with rivals Microsoft and Amazon, which hold 20% and 30% market share respectively.

Making Google Cloud a contender has come at a cost: Alphabet has already shocked Wall Street twice during quarterly earnings this year by proclaiming higher-than-expected capital spending due to the need to build more infrastructure to fulfill cloud demand.

“This is the moment Google Cloud was waiting for,” Dave McCarthy, who directs coverage of the cloud industry for research firm IDC. “A lot of the future growth at Alphabet is being looked at through its potential.”

In exclusive interviews with Reuters, senior Alphabet executives mapped out the gameplan that transformed the also-ran into a growth driver: a cultural shift to a more customer-driven sales approach; changing how it worked with rivals; and a renewed focus on delivering profits.

‘GET OUT OF HERE, KID’

Back when Kurian joined Alphabet, Google Cloud struggled to win enterprise customers, unlike the ads side of the business which was dealing with the biggest companies in the world.

“We would go to the ads team asking, ‘Hey, can you help us out with this customer?’ And they would basically be like, ‘Get out of here, kid,’” said Josh Gwyther, a startup founder who worked at Google Cloud from 2016 to 2025.

That’s not a problem anymore: Google’s scope of AI offerings have brought it into conversations with large companies that previously only considered Amazon and Microsoft. 

“The ads business is extremely healthy, but it’s not going to grow at the pace that we are,” said Matt Renner, Google Cloud’s president of global revenue. 

Nine of the 10 leading AI labs are now customers, Kurian said at an industry conference in September. They include OpenAI, Anthropic and Safe Superintelligence. Details of Google’s deals with two of those labs were first reported by Reuters earlier this year.

To get there, Kurian took a mallet to traditional practices, replacing it with what several employees called an “un-Googley” culture. 

Kurian, who left a job as one of Larry Ellison’s highest-paid lieutenants at Oracle, brought financial discipline to Google’s “loosely-run, ground-up culture” that encouraged side projects and experimentation, said Chirag Mehta, a tech analyst who worked at Google Cloud from 2017 to 2021.

To slash costs, Kurian opened new offices in cheaper locations, such as North Carolina and Poland. He audited Google Cloud’s contracts for internal services and renegotiated those where he deemed his unit was being overcharged by other departments, as was first reported in The Information.

He has ordered employees to focus on revenue rather than bookings. Google Cloud also shifted its sales strategy to target customers by sector rather than geography. That helped reduce sales reps being assigned accounts in industries where they lacked specialized knowledge, said Renner.

Its focus on generative AI has allowed it to catch up with rivals Microsoft and Amazon from a technical standpoint, according to some analysts. 

“We believe that the three clouds competitively are on roughly equal footing,” said Goldman Sachs managing director Eric Sheridan. “That’s a very different competitive positioning for Google Cloud now than two or three years ago.”

WORKING WITH THE ENEMY

For years, Alphabet had reserved the lion’s share of its own chips, or TPUs, for in-house use only. That changed in 2022, when Kurian successfully lobbied to move the group selling TPUs from Google’s core engineering unit into Google Cloud.

That move drastically increased Google Cloud’s allocation of TPUs because the unit could now freely offer the chips rather than having to receive approval from another part of the company, according to two people familiar with the move.

At a time when the world was scrambling for compute, Alphabet made the decision to make its chips available to not just its own DeepMind AI unit, but also to its competitors. 

That decision stoked tensions internally, according to former employees of both units, but it gave Kurian a stronger sales pitch for courting customers.

“We are the only hyperscaler with both silicon and models of our own,” he said. “How deep is your technical differentiation when the same stuff that you’re reselling can be bought from somebody else?”

Google Cloud quickly leveraged the opportunity to petition Anthropic to test TPUs as a viable alternative to Nvidia’s GPUs, according to a former Google Cloud executive involved in the partnership.

By 2024, Anthropic had seen enough to deploy Alphabet’s TPUs at scale. In October, it expanded its deal with Google to use as many as one million TPUs, worth tens of billions of dollars. The startup, which is now valued at $183 billion, has also tapped Amazon for chips as it reduces its dependency on Nvidia, which controls about 80% of the AI chip market.

“The whole world was sort of on GPUs, OpenAI in particular,” said Dan Rosenthal, who manages Anthropic’s partnerships with Google and Amazon. The need for chips “pushed us to be more flexible,” he said.

Other AI developers including Apple and Safe Superintelligence have since adopted Alphabet’s TPUs. Pichai told analysts on an earnings conference call this week that Alphabet is “investing in TPU capacity to meet the tremendous demand we are seeing from customers and partners.”

And although Google Cloud launched an enterprise version of its flagship model Gemini in October, Kurian told Reuters he would welcome adding OpenAI’s family of enterprise models to Google Cloud if the ChatGPT maker was interested.

POWER SHIFT

The ascent of Google Cloud is shifting the balance of power inside Alphabet. Current and former executives told Reuters Kurian has gained clout at Google’s weekly “leads,” the agenda-setting meetings where division leaders jostle over resources and priorities.

“What Thomas has been a powerful voice for is making sure that when we say we’re focusing on the user, that we’re focusing on the enterprise customer too,” Pichai said.

It still has a long way to catch up with its rivals and it will be expensive to get there. In July, Pichai hiked Alphabet’s projected capital spending for 2025 by $10 billion to $85 billion. This week, he raised the projection again to between $91 billion and $93 billion, adding that 2026 was likely to run an even larger bill.

With concerns over an AI bubble growing, Pichai told Reuters that he expects Google Cloud’s business to have “a lot of resilience” against short-term market corrections. “From our standpoint, we’ve been doing the AI thing for a decade now, and we’re going to be doing it a decade from now.”

(Reporting by Kenrick Cai in San Francisco. Editing by Kenneth Li and Michael Learmonth)