Google Cloud revenue jumped 82% YoY to USD 24.8B in Q2 2026
Gemini Enterprise now reaches nearly 90% of Fortune 100 firms
New TPU chips and a unified AI stack power Kurian's growth strategy
Thomas Kurian took over Google Cloud in 2019, when the business trailed far behind Amazon and Microsoft. Many expected him to spend years simply closing that gap.
He chose a different path instead. Kurian built Google Cloud around artificial intelligence rather than treating it as an add-on feature, and that decision now defines the entire company. Seven years later, the results speak for themselves, and rivals are watching closely.
Google Cloud's revenue growth has accelerated sharply in a short period. Revenue increased each quarter over the period shown.
Q2 2025: revenue grew 32%, reaching USD 13.6 billion
Q3 2025: revenue grew 34%, reaching USD 15.2 billion
Q4 2025: revenue grew 48%, reaching USD 17.7 billion
Q1 2026: revenue grew 63%, reaching USD 20 billion
Q2 2026: revenue grew 82%, reaching USD 24.8 billion
This is not a one-time spike driven by a single large deal. It is a sustained climb that puts Google Cloud ahead of Azure and AWS in growth rate over the period being compared, even though both rivals still generate higher total revenue.
The backlog tells an even bigger story. It reached USD 514 billion by mid-2026, up 390% from a year earlier, showing that companies are signing long-term AI contracts rather than running short trials.
Adoption numbers back up the revenue figures. Nearly 90% of Fortune 100 companies now use Gemini Enterprise in some part of their business. Gemini models handle about 22 billion API tokens every minute, a scale that was unthinkable for Google Cloud just a few years back.
Almost 1 billion people worldwide now use the Gemini app directly, which gives Google a consumer reach that few enterprise AI rivals can claim.
Also Read: Cloud Computing Trends Shaping Enterprise IT in 2026
Kurian's core idea is fairly simple to explain. Instead of selling cloud storage, computing power, and AI tools as separate products, Google Cloud bundles everything into one connected system.
Speaking at Google Cloud Next in April 2026, Kurian told the audience that businesses had moved past early testing. He said plainly that "the experimental phase is behind us" for most companies adopting AI. Scaling it properly, he added, needs "a unified stack" that ties together data, security, compute, and applications in one place.
That thinking led to the Gemini Enterprise Agent Platform, built to help teams create and manage AI agents across an organization. A few features stand out in this platform:
Agent-to-Agent Orchestration, which lets multiple agents work together on a task
Agent Gateway, which controls access and permissions for each agent
Agent Observability, which gives managers a clear view of agent activity
An Inbox tool inside Gemini Enterprise for tracking agent work day to day
Kurian explained the thinking behind it directly, saying the platform was built "to manage the entire lifecycle of an agent." That full-lifecycle approach sets Google apart from rivals still focused mainly on simple chat assistants.
Hardware plays a quiet but important role in this strategy. Google rolled out two new chips in 2026, the TPU 8t for training AI models and the TPU 8i for running them in real time.
Amin Vahdat, who oversees Google's infrastructure work, said the TPU 8t can reduce training times from months to just a few weeks. TPU 8i carries a larger memory cache and needs fewer network hops, which cuts delay for live applications. Alphabet plans to spend close to USD 200 billion on capital projects in 2026, with much of that money going toward chips and data centers.
Google still offers Nvidia GPUs alongside its own chips, so customers are not locked into one hardware choice. Kurian often points to this flexibility, covering both chips and AI models, as a key reason enterprises pick Google over more closed platforms.
No company adopts AI at scale without confidence in how it is governed. Google Cloud has built identity checks and compliance tools directly into its agent platform to address that concern.
Sundar Pichai has called this broader plan “mission control for the agentic enterprise,” a phrase that sums up how Google wants businesses to track AI work across every team. Demand for security tools has come up again and again on recent earnings calls, suggesting this focus is landing well with cautious buyers.
Also Read: Cloud Computing in 2027: A Complete Guide to Technology and Services
Thomas Kurian's strategy reflects a real change happening across the cloud industry right now. Companies no longer want to buy raw storage and computing power alone from their cloud provider.
They want a system that can reason, act, and automate work on their behalf. Google Cloud's results through 2026 show this bet is paying off, at least for the moment. Whether this growth holds steady as rivals sharpen their own AI tools is a question worth watching in the quarters ahead.
Also Read: 10 Fastest-Growing Cloud Computing Startups to Watch in 2026
1. When did Thomas Kurian become CEO of Google Cloud?
Kurian joined Google Cloud as CEO in 2019, arriving from Oracle. He chose to build the business around artificial intelligence from the start, rather than simply trying to catch up with AWS and Azure.
2. How fast is Google Cloud growing in 2026?
Google Cloud revenue grew 82% year over year in the second quarter of 2026, reaching USD 24.8 billion. This marked the fifth straight quarter where growth accelerated compared to the one before it.
3. What is Gemini Enterprise?
Gemini Enterprise is Google Cloud's platform for building and managing AI agents inside businesses. It covers orchestration, security, and tracking tools that support an agent's work from start to finish.
4. What are Google's TPU 8t and TPU 8i chips?
These are custom AI chips Google launched in 2026. TPU 8t speeds up model training, while TPU 8i is built for faster, low-delay AI responses when running models in real-world business use.
5. How large is Google Cloud's contract backlog?
Google Cloud's backlog reached USD 514 billion by mid-2026, up 390% from the year before. This figure represents signed contracts not yet booked as revenue, pointing to strong demand still ahead.