You’re receiving this email because you subscribed to the newsletter. If you’d like to unsubscribe, you can manage your preferences here with one click (or use the button at the bottom of the email).

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!

1. US government could buy a stake in Intel

We start this week with reports that the US government could buy a direct stake in Intel. The proposed deal would support Intel's domestic manufacturing ambitions, particularly its troubled Ohio fab that's been plagued by delays.

Intel’s CEO, Lip-Bu Tan, met with White House officials on 11 August to discuss concerns about the company’s direction. For over a decade now, it has failed to release new technologies on time and this has led to Intel being overtaken by its rivals.

If this move goes ahead, it would be an extraordinary intervention in the semiconductor industry. Following President Biden’s CHIPS Act in 2022, the US has been desperately trying to reduce its dependence on manufacturing in Taiwan.

It’s a continuation of the trend we’ve seen in recent years, with the US government viewing chip production as a matter of national security - rather than something that can be left to the free market.

2. NASA races to build a data pipeline to Mars

The agency is abandoning plans to build and operate its own relay spacecraft. Instead, NASA will outsource this job to private sector companies.

This shift has triggered a gold rush among aerospace companies, who are eager to become the internet service provider for Mars.

But of course, the stakes extend far beyond a single government contract. Whoever wins will control the data we need for space travel, as humanity focuses on the next chapter of exploration.

As it stands, there are four companies that hope to benefit - Blue Origin, Rocket Lab, SpaceX, and Lockheed Martin.

Blue Origin has already revealed its Mars Telecommunications Orbiter, which is aiming for a 2028 deployment. Rocket Lab is working on its own orbital relay concept, while SpaceX wants to adapt its Starlink constellation for Mars.

3. AI data centres are driving up electricity bills for ordinary Americans

Tech giants aren't just building data centres anymore. They're now setting up their own energy subsidiaries, investing in nuclear and fusion power, and then selling that electricity back to the utility companies.

The line between Silicon Valley and the traditional power industry is starting to blur in America, which is only going to accelerate in the coming years. In 2023, it was estimated that data centres consumed over 4% of the nation’s electricity.

But it’s believed that demand could triple and reach 12% in the next few years. This is mainly being caused by the AI boom, as significant amounts of energy are needed to train and run AI systems.

This explosion in demand has triggered a debate about who should foot the bill for massive grid upgrades, as ordinary people could see their electricity bills skyrocket. It’s believed that residents in Virginia will have to pay 25% more for electricity by 2030.

When we look at the wider picture across the US, it’s a similar story. Carnegie Mellon believes that bills will rise by 8% for everyone, which is bad news for Americans that are already dealing with a very high cost-of-living.

4. OpenAI is creating a startup to compete with Neuralink

OpenAI's venture arm could invest in Merge Labs, which is working on computer chips that work with the human brain. The funding discussions are still at an early stage, but sources have claimed that the startup could command a $850 million valuation.

It’s believed that Alex Blania is involved in the discussions, who currently leads a similar project that can scan your eye and verify that you’re human.

As I covered previously, Elon Musk’s Neuralink has been very successful with their human trials. Back in May, they showed how a nonverbal patient was able to edit and narrate YouTube videos - using only his brainwaves.

Investors were clearly impressed, as Neuralink was able to raise significantly more money. In 2023, it raised a pretty modest $43 million - but this increased to $600 million just 18 months later.

As a result, OpenAI is now looking into this area and how they can compete with Musk’s startup.

5. Anduril opens a rocket factory to boost Western militaries

The defence tech company has completed a new factory in Mississippi, which will be able to build rocket motors and compete with the industry’s giants - such as Northrop Grumman and L3Harris.

The facility aims to produce 6,000 tactical motors annually by 2026, with over 700 already passing the test firing phase. These motors will power everything from missile interceptors to deep-space probes.

It comes as Western militaries scramble to replenish their arsenals, following Russia's invasion of Ukraine. Europe is in particular need, as many nations have faced years of budget cuts and allowed their supplies to dwindle.

Earlier in the year, Anduril announced that it would build a $1.5 billion mega-factory in Ohio. This will allow the company to build their advanced drones and greatly ramp up production.

AI designs new antibiotics that can fight superbugs

MIT researchers have pushed AI drug discovery into uncharted territory and made a significant breakthrough. The team were able to use generative AI and design entirely new antibiotics from scratch, rather than simply screening existing compounds.

Their system has created two promising candidates, which have successfully killed drug-resistant gonorrhoea and MRSA in both lab and animal tests.

To do this, they trained their AI system on the molecular structures of 36 million compounds. It then used this data to understand how different atomic arrangements affect bacterial growth.

After that, they allowed it to design new molecules atom-by-atom. Two strategies were used, which resemble those used by DeepMind’s team a year ago on the AlphaGeometry project.

One strategy was to start with promising chemical fragments and then build outwards. While the other was to give the algorithm complete creative freedom.

Cleverly, the system was able to filter out anything that resembled existing antibiotics - allowing the researchers to only focus on new compounds. It’s fascinating to see the technology being used in this way and to deliver tangible outcomes for new medicine.

However, the researchers weren’t magically given the new antibiotic - they had to evaluate 80 theoretical treatments that were designed by the AI. Of those 80, only two could actually be developed.

It will be years before this new antibiotic is refined and used in human trials, but it offers real hope about how we can use technology to save lives. For years, we were promised that this was possible. Now, MIT’s researchers have reached that milestone.

China hosts its own “robot Olympics”

You might have seen some of these clips on social media, with robots competing against each other at running events and table tennis. Or, in one case, going off-course and crashing into humans.

China has launched its “World Humanoid Robot Games” and the event will feature 280 teams from 16 countries. These teams include a mixture of university and commercial companies, including China's Unitree, Fourier Intelligence, and Booster Robotics.

Crowds can enter the stadium and watch the events, with tickets ranging from $18-80. As you’d expect there were plenty of times when the robots collapsed and fell apart.

During a 1,500 metre race, one robot collapsed mid-sprint and prompted gasps from the crowd.

For the teams taking part, this offered them a chance to see their competitors’ robots and determine how their own could be improved.

It’s quite a light-hearted way to introduce small robots, which have a long way to go and are still in the early stages of development. But it also serves a strategic purpose for China.

The country has been pouring billions into humanoid robotics, as it faces an ageing workforce and growing competition with America.

The event also allows the Chinese public to become more comfortable with the technology and understand how it is progressing, as it will only become more widespread in the next decade.

🏭 SoftBank buys Foxconn's Ohio factory for the Stargate AI project

💰 Nvidia must pay 15% of Chinese chip sales to US Government

🚗 Ford revamps Henry Ford's assembly line to make low-cost EVs in America

🌌 Nvidia unveils new Cosmos world models and robotics tools

✈️ Joby Aviation successfully flies a robotaxi between two airports

💭 Sam Altman agrees that there's an AI bubble

🩺 Doctors that rely on AI actually become worse at detecting cancer

📱 ChatGPT's mobile app has already generated $2 billion

💧 UK government suggests citizens should delete files to save water

🔍 Perplexity offers to buy Google Chrome for $34 billion

While today's AI models can summarise a single video pretty well, they struggle when faced with thousands of hours of footage from multiple sources.

That means that marketing teams and security firms, which often have a huge number of video files, can’t use the technology on a larger scale and speed-up their work.

This is exactly what Memories is trying the tackle. The startup has developed a platform that gets rid of any noise in the video footage, compresses the footage, and only focuses on the essential information.

They then create search indexes, so that users can ask questions about their footage and get back accurate results. It works similar to Google’s search engine, except it has been trained to answer questions on your own videos.

Samsung is one of the key backers behind the startup, which has recently secured $8 million in seed funding. From Samsung’s perspective, they believe that the technology could be used to process videos on-device and address the privacy concerns around home security cameras.

Although I think the applications here are a lot more general. It could be used to remember your life through videos and quickly find the one you want. Or it could prove very useful for robotics, as they have to move around and remember where items were last placed.

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.