Google Dominate the AI Space with a Googol of Announcements.

📏 In a space where AI seems to be a dick-measuring contest, Google has apparently got some length. At their annual two-day flagship conference this week, Google Cloud Next, Google unveiled a swath of new innovations in AI, cloud infrastructure, and enterprise solutions.

From their seventh generation Ironwood TPU and new Gemini 2.5 model to Google Workspace Flows, an agent-to-agent (A2A) standard, Google went all out in their ​​commitment to advancing AI-driven transformation while strengthening its position in the competitive cloud market. (But more on that later…)

Fair to say that Google appears to be a grower in the AI race for domination.

🗞️ What we are covering today…

• Breakdown of Google Cloud Next 2025

• OpenAI countersues Elon Musk

• Amazon release a new foundation model, Nova Sonic

• Grok3 API drops with a few surprises

• Some cool new toys to use

• Jesus starts a podcast

Quick Note: There’s a lot to cover this week, and for a better reading experience, we suggest opening this in your browser!

🍾 FREE ENTRY TO OUR INVITE-ONLY AI CHAT ON TELEGRAM…

To celebrate our launch, if you share this newsletter and tag us on X, we will send you access to our invite-only Big Machines Telegram group, which is full of builders, investors, founders and creators.

This is a one-time offer, as the group loses its pazazz if anyone can join.

👁️ 👁️ What you might have missed

• If you missed Google Cloud Next 2025, then what rock have you been living under? A flurry of releases saw the tech giant dominate the AI space, which will be wrapped up below (you’re welcome).

• Just in time for WrestleMania, the other main event from the AI space sees OpenAI square up against Elon Musk, with the artificial intelligence giant countersuing its co-founder amid harassment claims.
  

• Amazon’s new Nova Sonic foundation model has made a pretty big leap in the AI space with the fact that it doesn’t just understand what you say, but how you’d say it. No longer will you be able to send passive aggressive requests to Alexa anymore without getting one back, though.

• ChatGPT just got memory across all your chats. This basically means it has context from all of your previous conversations. So, going forward, all of your info, quirks and how you like prompts answered, will continue to get better and better and be more relevant.
  
  This is a huge retention moat for OpenAI, which, during times of “Oh look, a new frontier model”, will work wonders due to the personalisation aspect.

• Grok3 API dropped at a lower context window than anticipated this week at 131,072. This is how it compares to Anthropic and OpenAI’s pricing.

🖥️ Google Dominate the AI Space with a Googol of Announcements.

Go big or go home. That’s what it felt like Google’s approach was to their two-day annual Google Cloud Next 2025 conference, where they released a dizzying amount of features with the complete list too long to cover in just a single newsletter. 

One of the most significant pieces of news was that Gemini 2.5 is now equipped with Deep Research, which is another company that uses the same name for its reasoning model. Does nobody do market research?

Just kidding, but it is confusing. Just skip to “Deepest Research” and be done with it. Anyway, it is pretty damn good at handling what its name suggests.

The Goog also followed in Amazon’s footsteps this week with a push towards reducing their dependency on hardware giants NVIDIA and AMD with their release of Ironwood TPU (7th Gen). This is a chip that is said to Power the Age of Inference.

DeepMind CEO Demis Hassabis revealed they are adding support for Antrhopic’s MCP to its Gemini models and SDK. This comes simultaneously with their Agent2Agent (A2A) standard, which is an open communication framework that allows any agent to talk to any agent to collaborate on overlapping tasks. 

So, what is the difference between MCP and A2A? We have you covered below…

The Gemini API has added two new entries, with Live API for real-time talking applications and VEO 2 for video generation. Gemini 2.5 Flash was also released with this low-latency model designed for tasks requiring real-time interference and conversations at scale.

Google Workspace also got an overhaul, with Docs getting audio overviews and Sheets getting AI analysis. Oh, and there’s a new automation creator, which they’ve named Workspace Flows. Now you can connect all your Google Apps with Gemini and Gems, using AI. About damn time. 

Another release that could spell trouble for the much-adored Replit and other IDEs is Google’s new Firebase. This full-stack app deployment IDE allows you to use text to create applications. Reactions have been pretty mixed, in all honesty, with some saying Replit, Lovable and Bolt get to fight another day. 

🙂‍↕️ Here’s everything we used this week. 

The winner of the one-shot hero, app-developing vibe-coder of the year goes too… not me; I am absolutely useless, but I did try a few new text-to-app models this week that are pretty sick. 

Rork - is a vibe coding platform that allows any old idiotic like you and me to create a mobile app in minutes. Given my wife is due to go into labour any day now and I am completely useless, I asked it to create a contractions timing app that would signal when the right time to go to the hospital is (smart, I know) - there are only 10,000 already in the app store, but that is beside the point.

Anyway, it was great. I highly recommend giving it a whirl. The fact that it is specifically for mobile apps is pretty cool, too. 

Higgsfield AI - Turns any image into a short motion graphic scene. It also allows you to choose from a variety of different scenes, including drone shots panning into your subject, some cool 90s fish eye scenes, and some stereotypical head-out-of-the-window shots.

It is worth trying out despite being a little fiddly during my first attempt. 

Reve - For any photographers reading this, it feels like VSCO, but for AI image generation, There is a feed of everyone sharing their work, and the best part is that you can click on each image and see the exact prompt that was used to get to the incredible images that are available.

It is a really good way to learn about prompt engineering when thinking about how to generate AI images, and it is also a cool place to take inspiration. 

Kairos - Alright, I haven’t actually got access to this one yet, but this video below shows how AI can be used with screen recording plus speech instructions to analyse your manual boring AF tasks and how an agent can create a workflow to replicate them which is pretty cool to me.

This is like training an intern or admin assistant once, and then it is automated forever. This one is for the micro-managers who never have to take over doing a task again… 

🤝 In Partnership with Mira Network.

🤯 Did you know that AI models hallucinate, and around 30% of their outputs could be false?

This creates significant challenges for anyone relying on AI systems for critical tasks. Mira Network has developed a solution to this problem with its trustless verification network that fundamentally transforms AI reliability.

Their decentralized approach creates a new paradigm for AI verification, enabling users to confidently rely on AI outputs across multiple models (DeepSeek R1, GPT4o, and Llama 3. 70B Instruct as of now).

If you are a developer, Mira's Public Testnet is now live at console.mira.network 

The future of AI isn't just about powerful models—it's about verified ones.

Don’t trust; verify.

The gloves are off: OpenAI countersues Elon Musk amid harassment claims

Elon Musk at the center of another controversy? Well colour us shocked.

The big bad billionaire has been hit with a countersuit by OpenAI – the company he co-founded – after being accused of creating a campaign of harassment to harm the artificial intelligence giant. 

The countersuit comes with an injunction to stop Musk from having any more interference with OpenAI, claiming he has “tried every tool available to harm” the company after alleging his $97.4billion offer to acquire them in February was not genuine and was a means to deliberately disrupt their fundraising efforts.

Plenty of shit has already been thrown but what does this mean for the future of AI? Who knows, but this battle of billionaires has only just begun, so stay tuned for how it’ll unfold. 

💸 Financials

• Ilya Sutskever, co-founder of OpenAI, had his on-paper net worth sent to new dizzying heights this week with reports that his new pre-product company, Safe Superintelligence Inc. (SSI), is valued at $32bn.
  
  Greenoaks have apparently led the round, according to the FT, whilst familiar faces of Alphabet, NVIDIA and Andreessen Horowitz are on the cap table, too.
  
  Sounds like a healthy market 🥹
  

• A16Z raised a megafund of $20 billion to invest in AI, with the funding going towards growth-stage investments in AI companies. The jaw-dropping sum, which could take several months to close, will reportedly be set aside for follow-on investments into AI companies included in the firm’s portfolio.
  

• KREA raised $83 million in series B, thanks to four contributors consisting of venture firms and technologists. The funding will go towards building their infrastructure, and researching multimodal AI systems capable of generating immersive environments in real-time.
  

• Runway raises $380 million in funding to expand its new media ecosystem. Following the latest round of funding, the startup is planning to develop AI research to focus on AI film using world simulators, with an aim to map out future events in that environment.
  

• Oracles announced a $5 billion investment over the next five years in UK cloud computing, supporting the government’s lofty ambition of becoming one of the leading players in AI software. The huge sum is set to enable Oracle Cloud Infrastructure’s footprint, along with helping industries like healthcare and manufacturing the new AI technologies in their businesses.

😮 Free shit!

• In the spirit of Oprah Winfrey, everybody gets an open-source MCP server! A complete collection of Model Context Protocol servers – over 300 listed and 100% Open Source – has been shared on X. These production-ready and experimental servers are now available for AI Agents.
  

• Can’t be arsed with coding? Well Google has released a new tool that can create AI agents that can do that all for you. A neat explainer video has dropped on X courtesy of Julian Goldie, who shows you how to implement these game-changing tools into your workflow.
  

• Christmas may be over eight months away but the guys at Notion have provided everyone with an early gift. Earlier this week, they made their MCP open-source, allowing you to plug it into your favorite client and start building richer AI integrations.
  

• It’s been quite a week for Google and alongside their many innovations, they’ve also published a 69(nice)-page whitepaper on Prompt Engineering and its best practices. If you’re using LLMs in production, make sure you don’t miss this explainer.

📆 The week ahead

• Alibaba is expected to release its flagship model, the Qwen 3, later this month. While an exact date is unknown, don’t be surprised if Alibaba gets it out sooner rather than later given the announcements in the AI world already in April.
  

• The Global Artificial Intelligence Summit and Awards (GAISA) 5.0 is taking place this weekend in New Delhi, India. AI leaders will be presenting ground-breaking research while recognizing the work in the field at this event.

👋 Until next week

What a week to start a newsletter on AI, eh? While trying to consume all this information is akin to the process of Foie gras, it’s hard to not get excited about where we are headed. To keep up with all the latest news and developments, make sure you’re plugged into this newsletter.

Better yet, if you REALLY enjoyed it, share it with a mate or your group chats, we would greatly appreciate it!

And around these parts, we always like to leave you with our meme of the week, so if you’ve ever wondered what Jesus Christ himself would look like on Joe Rogan, look no further: