💰 Inside Meta's $40 billion bet on AI

Welcome to this edition of Loop!

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

‏‏‎ ‎ HIGHLIGHTS ‏‏‎ ‎

• What Meta is spending $40 billion on and why

• Startup with $1 billion to discover new drugs with AI

• Anduril’s progress on autonomous aircraft

• … and much more

Let's jump in!

1. Meta will spend up to $40 billion this year on AI

We start with Meta, who held their Q1 earnings call last week and revealed plans to increase its investment in AI. Much of this money will be spent on new infrastructure and data centres, new chip designs, and ongoing research & development.

It’s a huge statement from Zuckerberg, who pivoted the company towards the “metaverse” just over 2 years ago. Speaking of their plans for Virtual Reality (VR), Meta confirmed that it was losing almost $1 billion on the project every month.

Adding all this with lower-than-expected revenues, and Zuckerberg’s warnings that their AI investment will take years to become profitable, have made investors nervous. Meta’s share price fell by 16% soon after.

It’s clear that Meta does not want to miss out on the next wave of AI and VR advances, like they did with mobile phones a decade ago. By not having their own mobile platform, they were forced to abide by Apple and Google’s clampdown on advertiser tracking - which is a major source of revenue for Meta.

Meta is pushing for that platform independence with the metaverse, but that still seems like a very distant future as the technology’s not there yet. In the meantime, the rise of GenAI could provide new revenue streams with advertising, but again that will take years.

The company has been very successful when it comes to AI research and has some of the best people working for them. Time will tell if this mega bet pays off, although I wouldn’t underestimate them.

2. AI drug discovery startup launches with a massive $1 billion

Xaira Therapeutics, an AI biotech company, has secured $1 billion in funding from prominent investors. The company aims to revolutionise drug discovery using their own advanced AI models.

These models, similar to those powering image generators like DALL-E and Midjourney, are designed to come up with molecular structures that can be created in the real-world.

Of course, this part of the industry is in its very early stages and there are lots of challenges ahead. It’s important to stress that the startup is likely to be using a mixture of tools, with GenAI used to suggest possible drugs and other software used to check if it’s actually feasible.

Google DeepMind have implemented a similar solution, albeit for a very different problem in mathematics, with their AlphaGeometry tool.

If implemented correctly, companies like Xaira could help us discover new drugs and also reduce the amount of money needed for researching them.

3. Eric Schmidt backs Augment to rival GitHub Copilot

Continuing the theme of big investments in AI startups, Augment has secured more than $252 million in funding to improve the productivity of software engineers. They’re backed by the former Google CEO and several big venture capital firms.

But Augment faces stiff competition from tech giants like Microsoft, Amazon, and Google - as well as many other startups. The market for AI coding assistants is becoming very crowded, which raises questions about how they will differentiate themselves versus everyone else.

The company has only emerged from stealth, with more details about pricing to be released later this year.

4. UK investigates Amazon and Microsoft over their AI partnerships

The UK’s Competition and Markets Authority (CMA) is starting a preliminary inquiry into these partnerships, as it wants to determine if they will negatively impact competition within the UK market.

It follows greater scrutiny of Big Tech's approach to mergers & acquisitions in the AI industry, with critics arguing that these partnerships are intentionally being used to avoid regulators.

Microsoft's investments in OpenAI, Inflection AI, and Mistral AI - along with Amazon's $4 billion investment in Anthropic - are under scrutiny from the CMA.

All of the companies involved have been asked to respond to their inquiry and provide them with evidence, before it decides whether further action is needed.

Separately, there are concerns in the UK that most big tech companies are not playing ball with their AI Safety Institute. Following the country’s AI Summit, all the major tech companies agreed to share their models with the government for safety checks - ahead of any public release.

This was a voluntary agreement, but only Google DeepMind has provided their models for testing. Other companies, like Anthropic and Meta, are in “active discussions” about how to work with the government.

There are obvious concerns here, as the companies are worried that it could set a precedent and lead to other countries getting prior access. That could soon become a problem, as these models are worth billions and are closely-guarded to prevent competitors from getting an advantage.

5. Apple’s latest AI model hints at what’s coming to the iPhone

Apple has released OpenELM, which stands for "Open-source Efficient Language Models". As the name implies, it’s designed to run more efficiently for text generation tasks - such as writing emails. The model is available in four sizes, ranging from 270 million to 3 billion parameters.

These lightweight models are even smaller than most AI models in their category, which is important as they’re cheaper to operate and can run on-device.

While Tim Cook has hinted that Apple is investing significantly in generative AI, we’re still waiting to see what that looks like.

We expect to learn more about their new AI features at Apple’s WWDC developer conference, which is scheduled for June, after one of their executives suggested it would be “Absolutely Incredible” (A.I.). Get it? Yeah, a bit on the nose with that one.

TikTok will be banned if ByteDance doesn’t sell the app

President Biden has signed a bill that will ban TikTok if ByteDance, its Chinese parent company, doesn’t sell it within a year.

ByteDance now has nine months to divest from TikTok, with a 90-day extension possible if they need more time to conclude a deal.

While both Democrats and Republicans supported this new law, it hasn’t been an easy decision for either party.

President Biden is running campaign videos on the platform, which are performing well and allow him to reach young voters. Similarly, senior Republicans are conflicted and don’t want to give up this online space to the other party.

Overall, the national security argument won out. There are genuine fears that this data is being sent back to Chinese servers, which can be accessed by their government. This data is incredibly sensitive, as it clearly shows the interests and motivations for millions of Americans.

That data is critical, as it allows foreign states to target them in the future. It also gives clear insights into their current political views, which is an absolute gold mine if you prefer one US party to win versus another.

The important thing now is to ensure that TikTok is sold to the right company - and kept out of the hands of people who already have a lot of control in the US media.

Synthesia’s AI avatars can now display emotions

Synthesia is a popular AI startup that allows you to create custom avatars for corporate videos. They’re mainly targeting customer service and marketing use cases for these avatars.

The company has recently upgraded their model, with it now able to create AI avatars that are more expressive and can show emotions based on your script. The results aren’t perfect and still look a bit weird, but there’s potential for this to become very realistic in a few years.

That’s both a great thing for businesses, as they get to create more marketing materials for little to no cost, and a problem since these avatars will inevitably be used to spread misinformation. I’m sure Synthesia will try to stop that, but it’s almost impossible to prevent.

According to the company, over 200,000 people have used the service and 18 million videos have been made. That’s an incredible figure.

At this year’s SXSW I met with the founder of AI Park, which is a startup based in Korea that also makes AI avatars. Surprisingly, they have been using these AI avatars to replace news anchors for a local TV station.

It’s strange to see that this technology is mostly here. Yes, the videos don’t always look perfect - but they’re not far off it. Only some small advances are needed to take it further, which could open a whole new can of worms for us all.

💬 Google releases a new AI feature to help people improve their English skills

💰 Elon Musk’s xAI is close to raising $6 billion in funding

📈 Google is officially a $2 trillion company

🤝 Meta will allow third-parties to make Quest VR headsets, such as Lenovo and Xbox

⚖️ FCC votes to reinstate net neutrality

📉 Tesla profits drop by 55%

🎵 Drake takes down his diss track that featured an AI Tupac

⛑️ Sam Altman, Satya Nadella, and other tech leaders join the US Safety Board for AI

Anduril

While they’re technically not a startup anymore, as they’re about 7 years old, I thought they were worth diving into - especially following their latest announcement.

Anduril has just secured a major contract to develop small, unmanned fighter jet prototypes for the US Air Force.

This is part of the US’ Collaborative Combat Aircraft (CCA) program, which aims to partner human pilots with unmanned aircraft. These unmanned systems will have their own weapons and 1,000 are scheduled to be delivered.

The defence company was started by Palmer Luckey, who’s just 31 years old and has previously founded Oculus - which was then sold to Facebook for $2 billion.

Anduril was valued at $8.5 billion during the last round of funding, which happened back in 2022. It has several other contracts with the US Government, including the Department for Homeland Security, Customs and Border Protection, the Marine Corps, and the UK’s Royal Navy.

They aim to shake-up the defence industry by using the Silicon Valley mindset of moving fast and incrementing, which contrasts with a slower approach you typically see in the industry. This means they will often develop products before the government expresses interest in them.

Given the people involved and contracts they’ve signed already, they’re certainly one to watch closely.

This Week’s Art

A lot has been covered this week, including:

• Why Meta will spend up to $40 billion this year on AI

• How a startup wants to discover new drugs with AI (…and $1 billion)

• Augment’s challenge to GitHub Copilot

• UK’s investigation into Amazon and Microsoft’s AI partnerships

• Apple’s very efficient AI model

• Why TikTok is seen as a security risk and must be sold

• Synthesia’s AI avatars that now show emotions

• And Anduril’s contract to develop autonomous aircraft for the US Air Force

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.

About the Author

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