♠️ A WEEK OF BIG BETS IN AI ♠️

Stop the world, we want to get off. That’s certainly what Elon Musk has been thinking, after he announced the mergers of companies SpaceX and xAI.

The company is now worth a supposed $1.25tn, making it the most valuable private company in the world today. Let that sink in for a minute.,

These record-breaking mind-boggling sums – from Anthropic’s latest $30bn valuation to some lunatic spending $70m on a single domain name – have been a pretty tidy theme for the past week.

And like everything, there are winners and losers in this game. Let’s get into it…

SpaceX acquires xAI 🚀 A poetic warning ⚠️ Anthropic’s February funding 🤑 ByteDance’s AI chip plans 🇫🇷 Marszalek’s domain name bet 🎬
Models of the week 🤖 Financials to watch 💹

P.S. There’s a lot in this newsletter – we suggest getting comfy and tucking in on desktop via the READ ONLINE tab at the top of this mailer.

THE WEEK IN AI

Our weekly news run has had a revamp. Get familiar with our news tags, so you can get straight to the point with the subjects that interest you the most
 

 .FRONTIER NEWSR.
#1 SpaceX’s xAI merger & co-founder conundrum
It should’ve been a week of celebration for Elon Musk. He’d just announced the $1.25 trillion merger between his xAI and SpaceX brands, revealing his intention to take the world’s most valuable private company public later in the year. But then xAI co-founders, Tony Wu and Jimmy Ba, exited the company. Push or pull? Musk insisted the high-profile departures were a push, and a symptom of aggressive scaling. “We’re re-organizing the company to be more effective at this scale,” Musk told the NYT. “When this happens, there’s some people who are better suited for the early stages of a company and less suited for the later stages.” Ouch.

It’s believed Wu and Ba feel differently, citing concerns about the new as-of-yet unnamed company’s growth prospects and alluding to its behemoth size as hindrance – “A small team armed with AIs can move mountains and redefine what’s possible,” Wu wrote in his exit memo. Six of xAI’s 12 co-founders remain. A raft of senior engineers have followed suite. All present the potential to become competitors. And it probably doesn’t get much cosier as controversy shrounding Musk is growing. The backlash over Grok’s deepfake sexualisation of women and children has intensified to a political level, while innuendo surrounding Musk’s appearance in the Epstein files swirls. “It’s just a distraction,” Musk claims.

Just a little, you’d think.

..TRANSFER NEWS ..
#2 “THE WORLD IS IN PERIL”
If dramatic exits are your thing, look no further than researcher Mrinank Sharma’s parting words from Anthropic. “The world is in peril,” Sharma lamented, after using his exit to cite concerns over frontier companies insistence insistence to pursue a growth-at-all-costs strategy. Sharma says he’ll be moving back to the UK to “become invisible” and “study poetry”. Maybe he’s seen too much after investigating issues such as AI sycophancy, bioweapon/bioterrorism risks, and dehumanisation of users (Read: Anthropic’s latest Sabotage Risk Report, cheery stuff). Maybe I can’t blame him. But poetry. Fucking poetry. Sharma is another high-profile AI exit this week, with OpenAI researcher Zoe Hitzig doing a runner from the Altman led firm, after criticising its decision to introduce advertising to ChatGPT. "Creating an economic engine that profits from encouraging these kinds of new relationships before we understand them is really dangerous," Hitzig said. Not quite as poetic; just as dramatic. 

 ..FRONTIER NEWS ...
#3 Anthropic’s valuation hits $350B
Anthropic just raised a massive $30 billion round, valuing the company at $350bn pre-money. A stacked investor list includes GIC, Coatue, Founders Fund, Nvidia, plus first-timers like Sequoia and Temasek. The Claude developer says the money will go towards infrastructure, hiring, and matching rivals like OpenAI and Google – not a bad idea considering all the money OpenAI could rake in from in-app advertising. Anthropic should rightly feel buoyant: Claude Code is a big hit with developers, and the company says it's pulling in around $14bn in annualised revenue, mostly from enterprise customers. An IPO could be on the cards this year, too; could we see a public offering-off to rival Musk’s? Meanwhile, OpenAI is reportedly eyeing its own round north of $100bn. The AI arms race rolls on.

 ..FRONTIER NEWS...
#4 ByteDance plans to develop its own AI chip
ByteDance is rumoured to be developing its first AI inference chip, codenamed SeedChip. And the Chinese firm is reportedly in talks with Samsung to manufacture it – something TikTok’s parent company denies. The company aims to receive samples by the end of March 2026 and produce 100,000–350,000 units this year. The Samsung deal would also secure increasingly scarce memory chips, and make the firm a major global player. It’s rumoured ByteDance plans to spend over $22 billion on AI procurement this year, reflecting a growing urgency among Chinese tech firms to develop in-house chips to sidestep US export controls. They better get a shift on then, as rivals Alibaba and Baidu are already well ahead in their own chip development.

 ..BILLIONAIRE BULLSHIT...
#5 Marszalek’s bets big on ai.com
Crypto.com founder Kris Marszalek has snapped up AI.com for $70 million – the most expensive domain sale ever – and paid entirely in crypto (well duh). The dot com acquisition was debuted during Super Bowl XL on February 8 – a spectacle dominated by casual racism and AI advertising – and purports to help “launch AI agents” or some bullshit. Apparently the ad’s mystery worked: it got so many visits after it aired it crashed the site (see above). The sale smashed the previous record held by CarInsurance.com at $49.7m, with broker Larry Fischer calling it a once-in-a-lifetime asset. Whether premium domains actually pay off is debatable, but Marszalek isn't exactly shy with spending – he’s already done his bollocks dropping $700m on stadium naming rights.

 ..MODEL BEHAVIOUR ..
#6 Deepfake fraud is going industrial
News just in: voucher scams are out, deepfake fraud is in. New analysis from the AI Incident Database believes the mode of fraud has gone industrial as tools virtually anyone can access and use are cheaper, more convincing and accessible than ever. No hacking expertise required, no Indian call centres. Recent examples include fake videos of Australia's WA premier promoting dodgy investments and AI-generated doctors flogging skin cream. A finance officer in Singapore was even duped into paying out nearly $500,000 on a fake video call with "company leadership," (how??) Researchers warn the worst is yet to come, with voice cloning already virtually indistinguishable, and video catching up fast. Now, it’s more than just your nana’s pension that could be drained – our trust in digital interactions is absolutely at stake.

..TRANSFER NEWS ..

#7 The Open in OpenClaw now makes sense…

OpenAI just snagged Peter Steinberger (the solo dev behind the viral open-source agent monster OpenClaw). He's joining to "drive the next generation of personal agents," per Sam Altman's X post, because apparently the future is "extremely multi-agent" or some bullshit. The project itself dodges full closure: it moves to an independent foundation as ongoing open-source that OpenAI will "support" (read: keep alive to look good while slurping the talent and ideas).

Announced Feb 15, right after the hype peak, OpenClaw exploded to the fastest-growing GitHub repo ever (199k+ stars), basically the AI that actually does shit like emails, bookings, insurer fights, while ChatGPT still chats. This is classic acqui-hire vibes: OpenAI gets Steinberger's multi-agent wizardry (and blocks Meta/Anthropic from poaching), boosts their agent game amid the race, and gets massive PR cred for "supporting open source." Whether it actually accelerates real personal agents or just bloats the roadmap is debatable, but Sam ain't shy about talent raids he's already dropped billions on compute and hires. Pure chadfish move in the agent wars, let’s hope OpenClaw stays open(ish), Peter gets infinite GPUs, and the timeline calls it Anthropic's biggest fumble yet.

TOOLS OF THE WEEK

The latest tool, model and LLM news, updates, tips and tricks – you name it, we’ll play with it til it breaks.

• ClawRouter: Looking to avoid “speedrunning your entire family into an underclass above for the goods. 👆” not, because you didn’t know what you were doing on OpenClaw? Fear not: our family over at Blocmates has produced this comprehensive guide on how to utilise the intelligence layer ClawRouter within the model. See
  

• Seedance 2.0: It’s not often we’re stopped in our tracks by a tool, and rarely are we convinced at developer claims that their tech will revolutionise creative industries yada yada. But Seedance 2.0, the ‘immersive audio-visual experience generator’ by ByteDance, is something else. Seeing is believing, folks.
  

• ChatCut: Currently in beta, this AI-powered video editor has been designed to edit videos by typing commands, eliminating the need for manual timeline manipulation. Join the waitlist today.
  

• TinyClaw: It may be the ‘year of the horse’ in the Chinese calendar, but for us, it’s very much been the year of the lobster*. Introducing TinyClaw – the lightweight wrapper for Claude Code, 99.3% smaller than OpenClaw. Check it below:

FINANCIALS

Covering significant funding stories, deals and partnerships, and investment opportunities

• Apptronik’s $520 million growth round, backed by strategic investors including Google and Mercedes-Benz, as signalled the growing commercialisation of physical/embodied AI. The funding will accelerate deployment of humanoid robots into industrial environments.
  

• Seligman Investments launched a new $500 million venture fund targetting AI, cloud infrastructure, cybersecurity, and data centres, indicating the continued institutional appetite for the physical and operational backbone of the AI economy.
  

• TRM Labs reached unicorn status with a $70 million Series C that valued the blockchain intelligence company at $1 billion, reflecting sustained demand for compliance and risk infrastructure as digital assets move further into regulated finance.
  

• Lawhive secured £43.9 million in Series B to expand its AI-native legal operating system for small businesses and consumers, underscoring continued investor interest in vertical AI for regulated professional services.
  

JOIN US ON TELEGRAM  📲

Share this newsletter with a friend and if they give us a follow you’ll gain access to our invite-only Big Machines Telegram group. Here you’ll find like-minded individuals such as builders, investors, founders, and creators.

Access is only granted to those who refer the newsletter to a new subscriber, so if you try sign-up with a work email (this tanks our open rate!), we’ll know, you sneaky bastard.

GO ON, GIVE US A SHARE…  

You currently have X referrals.

UNTIL NEXT WEEK, FRIENDS.

To close the week, we’ve been following the uproar on socials over OpenAI’s plans to sunset GPT-4o (check the replies 😩). Life-through-the-looking-glass stuff.

Still, we’re pleased to see your dad is taking it well…