Google Cloud Next 2025 — all the news and announcements as they happened
Google Cloud Next 2025 is over - here's what we saw
Good morning from sunny Las Vegas! We're here for Google Cloud Next, and after a slighted delayed arrival into the city last night, are busy getting ready for the event kicking off tomorrow.
Before the conference kicks off tomorrow, we've been invited for a VIP press conference at an undisclosed location in Las Vegas...
Well, that was interesting - we were whisked off to the famous Las Vegas Sphere, for a look at how Google Cloud worked to help prepare the movie classic The Wizard of Oz for the venue, launching on August 28.
The event was introduced by Alphabet and Google CEO Sundar Pichai, who welcomed us to the show, and outlined the scale of the technology being used.
Make no mistake - seeing a 75-year-old film, originally filmed in a rectangular format, expanded and spread across the screen, is pretty spectacular...
The event then wrapped up with a discussion between Google Cloud CEO Thomas Kurian and Jim Dolan, CEO of Sphere Entertainment Co, covering how the partnership came about, and their hopes for AI in the future.
Kurian then took to the stage solo to welcome us to Google Cloud Next 25.
Good morning and welcome to the first day of Google Cloud Next!
We're off for breakfast and caffeine, before the keynote starts at 9am PST.
We're en route to the keynote now - as you might expect, it's pretty busy...
We're in and seated! As per Vegas conference rules, we're being blasted with upbeat techno - accompanied with strange and fantastic animated scenes generated by Google's Veo2 model...
Brass instruments floating over the Las Vegas strip and the Grand Canyon? It's amazing what AI can do...
Here we go - the lights go down and it's time for an intro video, again showing some...interesting visuals generated by Veo 2.
Google Cloud CEO Thomas Kurian takes to the stage.
Our first special guest - Sundar Pichai, CEO of parent company Alphabet.
Pichai adds that Alphabet is ramping up investment in AI infrastructure - and reveals the first news of the day - Google is making its Cloud Wide Area Network open to businesses across the globe.
That's not all - Pichai unveils Ironwood, the company's next-generation TPU.
Moving quickly on - Pichai outlines the success of Gemini 2.5 Pro, which he says is streaking ahead of the competition.
Pichai reveals Gemini 2.5 Pro is now available for everyone in AI Studio, Vertex AI, and the Gemini app.
There's also the release of Gemini 2.5 Flash, offering even more access to the platform.
There's also a mention for Veo 2, which is being used by advertising and marketing agencies across the world, Pichai says.
We're then treated to a video showing how McDonald's is using Google's AI tools, using real-time data to improve performance and boost the overall restaurant experience - tasty stuff.
Kurian returns to the stage, highlighting why customers want to work with Google Cloud.
We welcome Amin Vahdat, VP and GM, ML, Systems and Cloud AI, to the stage, to talk about hardware.
This will be backed up by Nvidia GB200 Blackwell chip-powered hardware - and Google will also offer Vera Rubin GPUS on Google Cloud.
Vahdat also runs through a host of new software announcements, including boosts for GKE Inference Gateway, Pathways on Google Cloud, and vLLM to TPU, giving easy TPU inference on a familiar library.
"We're truly seeing tremendous momentum," Vahdat notes, with Google helping power AI unicorns across the globe, as well as major customer usage across a wide variety of industries.
Google Distributed Cloud also gets an upgrade, with new Gemini tools allowing Google Distributed Cloud to run locally in air-gapped and connected envrionments.
A quick video from Nvidia CEO Jensen Huang is up next, highlighting the partnership between the two.
Vahdat moves on to Google Workspace, which is getting a major boost from Gemini, which is now included in all subscriptions.
Imagen 3 is next up, now offering greater image generation than ever, along with voice-generation tool Chirp 3.
Veo 2 is also much more powerful than ever, offering first and last shot control, more camera angles, extra inpainting and outpainting tools, and much more.
After that bombardment of information, it's time for a demo of just what Vertex AI can do.
Kurian returns, and it's time to talk about Vertex AI.
Vertex allows users to discover, customize and deploy the best models for your business, he notes, introducing a customer example from Intuit, who used AI to help even the most complex tax filings a breeze.
Tens of thousands of companies are using Vertex to find the best models, Kurian notes - with use cases from patient healthcare to energy management.
Agents are never far away, with the tools now able to be easily built on Vertex, connecting to a number of top third-party partners.
Next up, Kurian wants to talk more about agents.
Google Cloud will now let users build and manage multi-agent systems, Kurian reveals.
There's also a new Agent2Agent protocol to let agents talk to each other, regardless of what model they were built on.
We're now seeing how Agentspace works in action, with a demo showing off how agents can be used to automate everyday tasks.
Agentspace is now integrated into the Chrome browser, Kurian reveals, meaning you can search company data directly within your browser.
We're going even more in=depth on agents now, taking a look at several key areas where the technology in making a difference.
Next are some new offerings to improve Agents via Customer Engagement Suite, including streaming video support, new voices, and AI assistance to build agents.
Industry agents are also a key use case, no matter how specific.
Creative Agents are another key area, augmenting creative teams to allow content production at scale, or reimagining existing content.
Agencies across the world are using Google's models to help create ads, improve localization, and reduce production times across the board.
Data agents are also set to be a major boost for customers, taking lots of the heavy lifting and time-intensive tasks away from busy workers.
Google's BigQuery platform offers more reach than any competitor, the company says, bringing together structured and unstructured data to build the most useful agents for your business.
Next up, we're looking at Code Agents - as Gemini can be a highly effective partner using its Code Assist tools.
Today new agents are being launched for Google Code Assist, including greater insights on the task being worked on, and more interactivity options.
Now, we're moving on to security, with Sandra Joyce, VP of Threat Intelligence.
A big one - Google Unified Security is a new converged security solution bringing together all of the company's tools and features.
GUS looks set to revolutionize how companies ensure they stay protected, offering real-time advice and recommendations defined to even the smallest threats...
We're nearly at the end, and Kurian returns to the stage to wrap things up.
"What an amazing time for all of us to experience and work with these technology advances."
"We are honoured to be building this new way to cloud, with all of you."
And that's a wrap! We're off to digest all of the mountain of new announcements made in that keynote, so stay tuned to TechRadar Pro for write-ups of the biggest announcements coming soon.
We've got our heads around some of the news from this morning, so if you missed it - here's some of the biggest announcements...
The company has unveiled Google Unified Security, a new unified security platform designed to take the stress out of keeping your business safe from the latest threats.
Google Unified Security, affectionately known as GUS, promises a major step forward in threat detection and mitigation, and is outfitted with the company's latest Threat Intelligence, AI tools and services to keep users safe.
Google unveils new security AI agents to keep your business safe from the latest threats
First is an Alert Triage agent in Google Security Operations which is able to carry out dynamic investigations on behalf of users, reducing the workload of analysts who otherwise are triaging and investigating hundreds of alerts per day.
The other new addition is a Malware Analysis agent, part of Google Threat Intelligence, which can investigate whether code is safe or harmful. It will also have the ability to create and execute scripts for deobfuscation, before summarizing its work and offering up a final verdict.
Google Cloud unveils Ironwood, its 7th Gen TPU to help boost AI performance and inference
At its top-end Ironwood can scale up to 9,216 chips per pod, for a total of 42.5 exaflops - more than 24x the compute power of El Capitan, the world's current largest supercomputer.
Each individual chip offers peak compute of 4,614 TFLOPs, what the company says is a huge leap forward in capacity and capability.
We also heard from Alphabet and Google CEO Sundar Pichai at the keynote, who reaffirmed the company's plans to spend big on infrastructure and technology going forward.
And that's a wrap on our live coverage from day one of Google Cloud Next 25 - it's been a busy day, so we're off to rest and recuperate, before heading back tomorrow!
Good morning and welcome to day two of Google Cloud Next 25!
First up, we have a media Q&A with Thomas Kurian and others, before the developer keynote later today, so stay tuned for all the news from those events and more!
First up today, we have a roundtable with some pretty big names, including Google Cloud CEO Thomas Kurian, alongside Tara Brady, President, Google Cloud EMEA, and Eduardo Lopez, Vice President, Google Cloud LATAM.
The panel is here, and we have some introductory comments from Kurian on what's been the most important announcements from the show for him.
"Our success is reflected not by what we do, but what our customers do with our technology," Kurian notes, shouting out some customer success stories from the EMEA and LATAM regions.
Brady notes the growth has been "phenomenal" in EMEA over the past year.
The first question is about how Google Cloud deals with complex geo-political situations, and if this affects its business outlook.
He adds its Sovereign AI systems are controlled in a local environment, managed by a national foundation - and that the company works hard on localizing its products for different countries.
Asked about the EMEA region, Brady remarks that security is a crucial consideration, along with flexibility - and of course AI, and the competitive advantage this offers.
Next is a question about how companies like Google Cloud can cope with the energy demands caused by AI expansion.
"You'll see us continue to optimize the cost of serving models, while improving quality," he notes.
A pretty staggering stat from Kurian - Google Cloud has seven times the amount of water-cooling systems than the entire rest of the world, combined.
"For us, it's really important people see AI as a key technology in driving the consumption of clean energy," he says.
Unsurprisingly, a question about the US tariffs is next - how will Google Cloud look to deal with the possible effects?
Kurian is asked about UAI tools - he replies the entire approach is about simplifying technology for users everywhere.
"You have the same technology that runs Google, available to every small business...by simplifying the technology, you bring it to everybody."
Next is a question about Agentspace, and how it will affect companies which already have a presence with other ecosystems.
The next question is about building new data centers - how does Google Cloud decide where its next projects will be?
"You'll see us continue to expand our global footprint," he says.
Another question on tariffs - and if this will lead to national clouds in the future, especially if the US (or other countries) suddenly decide to cut off access?
"The proof is in how many customers are choosing to run (Google Cloud)".
The final question is about Google's proposed acquisition of Wiz - is it really worth that much?
And that's a wrap on the Q&A - we're off to get some coffee, but we'll be back for the developer keynote later today.
We're back, and headed to the day two keynotes - this one is for the developers, so expect a deeper dive into some of the major announcements we saw yesterday - and possibly some extra surprises, too...
We're nearly there - but going to be another packed turn-out, it looks like...
We're in and seated - and it's time to kick off!
The lights go down, and Brad Calder, VP & GM, Google Cloud, takes to the stage.
Its new Agent Development Kit, Agent Engine and Agentspace all get an early shout-out.
"Together, these allow you to create agents where AI and humans can work together to reach a common goal."
All of this, of course - is powered by Gemini.
From getting started, to scaling, to a sneak peek at the future, we're about to take a look at a lot of demos...
Our first demo looks at how you can build AI-enabled apps with Gemini in AI Studio.
As the model thinks, we're given a run-through of exactly what is going on behind the scenes - including the in-depth "Thinking" box.
The end result is a beautifully-designed new kitchen idea, which can then be customized even further, and take in the latest Google search data for real-time details on information such as building material costs.
Gemini's long-context windows are getting a shout-out, giving users much more ability and flexibility to create incredibly detailed prompts, and outputs.
Next, we're moving on to how agents can help businesses themselves.
Our next demo is of Agent Development Kit - we're going to see just how easy it is to build an agent.
The demo shows how easy it is to connect to different models, using natural language instructions to carry out its task.
With all that done, a final document is created in seconds - a huge time-saver for users everywhere.
Next up, we're looking at how developers are building with Vertex AI Agent Engine, now generally available.
We're being shown a demo of how easy it is to build even complex, multi-source systems in ADK.
The demo then shows how straightforward debugging a troublesome agent can be - with just a few steps to get the tool up and running again.
Next, we're moving on to how Gemini can be useful in IDEs and tools for developers.
The company is allowing devs to use Gemini wherever they create code, and we're then taken to a look at how different models for different use cases can be accessed.
Now, we're taking a look at a customer use case, between Google Cloud and Major League Baseball.
We're now moving towards what's next for Google Cloud...first up, the next generation (already) of data agents.
Our next shift is to look at the future of software development through agents.
Code Assist is getting way more customization options and capabilities, giving developers all kinds of ways to tailor apps exactly how they want them.
And that's a wrap on the day two keynote - and for Google Cloud Next 25 as a whole.
Stay tuned for more of our wrap-up coverage, but thanks for joining us over the last few days, and thanks for reading TechRadar Pro!