Welcome to this edition of Loop!

To kick off your week, I’ve rounded-up the most important technology and AI updates that you should know about.

‏‏‎ ‎ HIGHLIGHTS ‏‏‎ ‎

Let's jump in!

Ad algorithms reward volume, but most teams can’t keep up without sacrificing either quality or brand. Hightouch changes that.

With Ad Studio, performance marketers can turn their data into hundreds of high-quality ads in minutes. Every asset is grounded in real insights, built for easy editing, and ready to launch — across any platform.

React to performance in real time. Jump on cultural trends while they’re still relevant. Respond to competitor campaigns before they gain traction.

More output, better inputs, no trade-offs required.

See Ad Studio in action

1. Google's new AI model that can “create anything”

We start this week with Google's launch of Gemini Omni Flash, which is different to most AI models as it allows you to create anything from any input. For example, you could upload a photo of your dog, hum a tune into your microphone, and ask it to generate a video of your dog dancing to the song.

The company is gradually rolling it out and expanding features over time, so you can only generate and edit videos right now - but they plan to do the same for image and audio shortly.

While creating videos of your dog dancing is pretty funny, it's a novelty that wears off quickly. Instead, the real value here is that Omni flips video creation from something that's expensive and requires specialist skills, to something that's cheap and easy for anyone to do.

This could be really helpful for small businesses, as they can use this AI model to produce their own marketing videos and better advertise their products. And as social media platforms continue to prioritise video clips over every other form of content, that's incredibly useful for those who never had the budget or skills for this kind of work.

It's also important that platforms and users are told when a video was created with AI, so Google's SynthID watermark will be built into any videos that are generated. For some time now, Google's watermarking tech has been the best in the industry and one that I've regularly flagged in my newsletter - with OpenAI now agreeing to use the technology for their own platform as well.

If you want to find out more about the new model and what it can do, I've included a link below.

2. Europe's plan to build AI-powered military satellites

Over in Europe, two of the continent's biggest defence companies are planning to build military satellites that are powered by AI. The joint venture is called KIRK, and it's being led by Helsing (Europe's leading AI defence company) and OHB (one of Germany's biggest space groups).

In practice, this could be a big change for how European militaries operate. Instead of having to rely on US satellites for targeting data, countries like Germany, Norway and the UK would have their own AI-driven system in orbit.

These satellites would be able to spot targets autonomously and pass the coordinates directly to weapons systems, which significantly cuts down the time between seeing a threat and acting on it.

This is part of a broader trend, as European countries try to rebuild their militaries and counter a growing threat from Russia - while also moving to reduce their reliance on American technology.

European leaders have held crisis talks several times over the last six months, as a potential US invasion of Greenland has shown that the continent is too reliant on American tech.

With this move from Helsing, they're trying to give European militaries more of a homegrown alternative and greater tech sovereignty.

3. Meta cuts 8,000 jobs to pay for its AI spending

Meta has announced that it's laying off around 8,000 employees, which is roughly 10% of its workforce. The cuts come as the company tries to offset its enormous AI spending.

Back in January, Meta forecast that it would spend between $115 and $135 billion in capital expenditure this year - which is almost double the $72 billion it spent in 2025.

In an internal email, management told staff that the layoffs were part of an effort to "run the company more efficiently and to allow us to offset the other investments we're making." Alongside the job cuts, Meta also plans to move 7,000 staff onto new AI initiatives and will be closing 6,000 open roles.

We often talk about the hundreds of billions being poured into data centres, chips and energy as the cost of the AI race, but companies are now being explicit that job cuts are being used to pay for this.

4. Nvidia's next $200 billion bet is on CPUs

Nvidia reported its latest earnings this week, with another record quarter that brought in $81.6 billion in revenue. Around $75.2 billion of that came from data centres alone, which is up 92% from this time last year.

For context, Nvidia's revenues were just $5.66 billion in the same quarter five years ago - with data centres bringing in only $2.05 billion. The company has seen tremendous growth with the rise of Generative AI, as its quarterly revenue has increased by more than 14x.

But the more interesting part of the earnings call was Jensen Huang's pitch for the next phase of growth. Huang spent a lot of time talking up Nvidia's new Vera CPU, which he claims opens "a brand new $200 billion market" for the company.

It's quite a move, as Nvidia has historically dominated the GPU market instead - while Intel and AMD have focused on CPUs. Huang's argument is that AI agents - the next big wave of compute demand - mostly run on CPUs rather than GPUs.

So as agents scale into the billions, each one will need a CPU that's specifically designed to process tokens quickly - which is what Vera is designed for. It seems that the demand is real, as Nvidia has already sold $20 billion of these chips this year.

This is all happening at a time when the AI hardware boom is lifting almost every chip maker. Even Apple has seen its Mac Minis fly off the shelves, as developers and small AI labs snap them up to run models locally.

5. Figma takes on Claude Design with a new agent

The company has launched a new AI design agent that lives directly on your design canvas - allowing teams to quickly iterate and test new ideas.

If you're not aware of Figma, it's a popular design tool and most product teams use it to visualise new apps and websites before they're built.

This new AI agent is able to understand your own design system, so it can use this context to improve the results and stay aligned with your brand.

You can also run several agents at the same time and ask each agent to explore a different design direction. This means that you can compare three or four ideas side-by-side, rather than having to work through them one at a time.

The launch follows Anthropic's release of Claude Design last month, which can also turn text prompts into prototypes, slide decks and UI mockups.

Anthropic's tool has a slightly different target audience, as it's aimed at founders and product managers that don't have a design background. Whereas, Figma is built for designers and allows them to accelerate their own work within the platform.

When Claude Design was launched, Figma's stock immediately fell by around 6% - so this design agent is their response to that new competition.

Google launches a 24/7 agent that lives in your inbox

Google has announced Gemini Spark, which is a 24/7 personal assistant that runs in the cloud and can work on tasks while you're away from your computer.

Unlike a typical chatbot, which just answers questions, Spark is built to actually take actions on your behalf. And since it runs on a dedicated virtual machine in Google Cloud, you don't need to leave your laptop open or check back in for it to keep working.

The interesting part of this launch is the integration. Spark is being launched with connections to Gmail, Docs, Sheets, Slides and the rest of Google Workspace - so it can use whatever it finds across your files to get its work done. You can even email Spark directly and ask it to do something for you.

In Google's demos, Spark was able to scan information from a list of emails and documents - then draft a status update that could be sent to your boss. It was also used by a small business to monitor their inbox, then flagged questions from customers as soon as they came in.

It's worth looking at how this compares to Anthropic's Claude Cowork, which launched late last year. Cowork is Anthropic's version of the same idea - running tasks in the background and connecting to your tools - but it sits inside Anthropic's environment rather than inside Google Workspace.

Spark's advantage is that your emails, docs and sheets are already there, so the agent doesn't have to be granted permission to a long list of third-party apps. Of course, when you have Spark doing real work for you across your inbox and documents, it might become harder for you to leave the Google ecosystem and try other AI tools.

It's also worth flagging the upfront cost here. This agent is only available on AI Ultra plan, which starts at $100 per month and is one of the company's most expensive tiers.

There's also a broader trend here, as tech companies begin to reserve their most advanced AI features for the highest subscription tiers. This is being done to claw back some of the heavy losses they're seeing with AI inference and I expect to see a lot more of this in the coming year.

If you want to read more about what Gemini Spark can do, I've included a link to it below.

⚖️ California Governor signs executive order to prepare for AI job losses

🎓 Students boo speakers for promoting AI at university graduation, as the job market becomes more difficult

💰 Anthropic says it's about to have its first profitable quarter

🔄 OpenAI co-founder Andrej Karpathy joins Anthropic

👨‍⚖️ Elon Musk loses his case against Sam Altman

🤝 Anthropic is paying $15 billion a year to use Elon Musk's data centres

🕵️ Google is making it easier to detect deepfakes in Search and Chrome

🗺️ World models allow you to simulate real places within Street View

🚫 ArXiv will ban researchers who upload papers with AI slop

🚗 Wayve is bringing its self-driving tech to the US, with cars made by Stellantis

🎵 Stability AI's new audio model can create 6-minute songs

🔖 OpenAI now supports Google's SynthID standard for watermarking AI content

This startup has developed AI agents that can detect and stop cyberattacks in real-time, taking a lot of the burden off human security analysts.

Exaforce was founded three years ago and they've just raised a $125 million Series B at a $725 million valuation. Overall, they've managed to raise $200 million in funding from investors.

They've developed their own platform that can triage alerts, run investigations and automate some of the work that analysts do. The product was only launched in Q4 last year, but they've already signed 20 customers, including Replit and Guardant Health, and expect to hit 40-50 by year end.

Interestingly, they have a feature that they call "vibe hunting" which allows security teams to query the platform with natural language. For example, you can ask "Did we get any new attacks from Iran?" and their AI agents will then investigate and search through your logs.

For the last few weeks, there have been several cases where AI models have spotted vulnerabilities faster than human teams can respond. While we can use the same technology to try and defend against these attacks, security teams can quickly become bogged down with alerts that are often false positives.

Exaforce seems to be trying to tackle this problem and allow teams to focus on the alerts that actually matter. It's great to see this kind of investment in cybersecurity, as the sector has found it harder to raise money in recent years compared to the new wave of AI labs.

This Week’s Art

We’ve covered quite a bit this week, including:

If you found something interesting in this week’s edition, please feel free to share this newsletter with your colleagues.

Or if you’re interested in chatting with me about the above, simply reply to this email and I’ll get back to you.

Have a good week!

Liam

Share with Others

If you found something interesting in this week’s edition, feel free to share this newsletter with your colleagues.

Liam McCormick is a Senior AI Engineer and works within Kainos' Innovation team. He identifies business value in emerging technologies, implements them, and then shares these insights with others.