📰 Welcome back!

We’re back after a brief pause, but we think this bumper edition will have been worth the wait.

Yes, the AI Bubble has dominated the news cycle, but with every day passing it’s becoming increasingly unclear what the actual fuck will happen. On one hand we have the US treasury dismissing claims of a bubble, but also ruling out the potential for a bailout. On the other, we have Anthropic and Alphabet’s CEOs telling us to be wary of the tech’s power and precarious market position.

Like gauging when a snot bubble will burst, we’re more worried about what’s to hand to clean it up – tissue or shirt sleeve. And as we the people are the ‘sleeve’ in this analogy, we better hope somebody has a plan to protect and insulate from the potential shitty mess it could bring.

It’s not all doom and gloom, though; we’ve got Matthew McConaughey speaking Spanish thanks to AI. Está bien, está bien, está bien…

🚀 What we’re covering today…

• ❌ No AI bailout from US government

• 🚪 Intel loses top exec to OpenAI

• 🗣️ Hollywood stars sign AI voice deal

• 🤫 Is Meta’s chief AI scientist about to bail?

• ‼️ AI CEOs offer stark warnings

• 🔥 Bezos launches Project Prometheus

• 💰 Elon’s $1 trillion Tesla pay package

• 📚 ‘Vibe Coding’ is Collins’ Word of the Year

• 🦾 Robots perform surgical first

• 🍎 Apple finalize Google Gemini deal 

• 🍽 LLMs are encouraging eating disorders

• 👑 The UK’s north gets huge AI investment

• 🤖 OpenAI releases ChatGPT 5.1

🔴 Quick Note: We like to cover loads of AI news in our newsletter, so for a better reading experience, we suggest opening this in your browser for the full experience! 

Head to the ‘READ ONLINE’ tab at the top of this email.

👁️ 👁️ What you might have missed

• We open this edition’s newsletter with the launch of a new franchise: ‘Blowing Bubbles’ – or for the metaphorically challenged, ‘It’s definitely not a bubble’. Now over a week ago, The White House declared it will not provide a bailout to the artificial intelligence sector, a move that will have Altman and team’s bumholes clenching tighter than a four-year-old gripping a crayon. It signals a serious concern across the political spectrum about over-investment in the sector, and this massive bubble is ready to burst and shower the global economy in shit.
  
  It seems the waters about who foots the bill for exponential investment have been muddied by lose lips in the industry’s upper echelons. Sentiment in the private sector seems to be that to support infrastructure growth, “an ecosystem of banks, private equity, maybe even governmental” support will be needed to “backstop” investment against currently modest returns (OpenAI CFO Sarah Friar’s words, not ours), and this has been a key point of consternation for critics of tech companies who believe “they’re trying to grow too big to fail”.
  
  US AI and crypto czar, David Sacks, emphasized that the market, not taxpayers, should handle any failure of big AI players, given that there are “at least five major frontier model companies”, ready to step in lead the way. Altman himself agreed, saying “taxpayers should not bail out companies that make bad business decisions or otherwise lose in the market. If we get it wrong, that’s on us,” albeit with all the confidence of said crayon-wielding child who’s just discovered reverse psychology.
  
  Other industry leaders are joining to the conversation, but it remains to be seen which way the US government – and indeed other major world economies – will go, to prevent an even worse global economic meltdown caused by the 2008 Financial Crisis.
  
  

  

  Sam Altman, starring in 2025 AI bio comedy drama, The Big Shit.
  Source: We made it/remaker.ai

• Having said this, it hasn’t been a bad couple of weeks for OpenAI, tbh: they’ve only gone and bagged Intel’s chief technology and AI officer, Sachin Katti. Katti, who led Intel’s efforts to develop AI chips and software rivalling Nvidia and AMD, will work on OpenAI’s compute infrastructure. It’s another high-level exit at the chipmaker, following departures of execs including Justin Hotard and Safroadu Yeboah-Amankwah. Despite recent cash injections from the Trump administration, major tech players such as Intel and SoftBank are lagging behind in AI innovation. And what will feel like daylight theft for the Santa Clara giant, has been compounded by actual theft, after a former Intel engineer has been accused of stealing over 18,000 files on their way out. Ouch.
  

• It probably has been a good couple of weeks to heed the warnings of the billionaire CEO class, however: Anthropic CEO, Dario Amodei, has warned that AI companies must be transparent about the risks of their technology, or else they risk repeating the mistakes of the tobacco and opioid industries, which concealed harm. Alphabet CEO Sundar Pichai, meanwhile, has cautioned against blindly trusting AI tools, stating they can be “prone to errors” and urging users to double-check their outputs. Maybe tell us something we didn’t know – Grok has briefly claimed Donald Trump won the 2020 presidential election, while Grokipedia is indulging white supremacist fantasies. What’s next? Pichai saying no company – not even Google – would be safe from an AI bubble bursting? Oh wait, he did exactly that.

• Jeff Bezos is stepping back into an operational role within his fiefdom, as co-CEO of a new AI startup called Project Prometheus, alongside physicist-che­mist Vik Bajaj, formerly of Google X. The stealthy company has already raised $6.2 billion, making it one of the most well-funded early-stage AI ventures. The project will apply AI to engineering and manufacturing processes across sectors like computing, automobiles, and aerospace, and his backed by nearly 100 staff that have been hired from across OpenAI, DeepMind, and Meta. As a gauge for how effective the project could be, it’s bemused Musk, who’s teased Bezos of “copycatting” his ideas. Talking of borrowing ideas, while Prometheus was a thief, the project’s big unveiling was supposedly not a spoof of Apple’s 1984 launch advert – get a load of this…
  

  
  

  

  Not Big Brother, seriously. Source: Alexander Tamargo/Getty Images.

  
  
  

• Alright, alright, alright: Hollywood royalty Matthew McConaughey and Michael Caine have each signed on with AI audio company ElevenLabs to license their voices for synthetic-audio use – something we clearly need more of, given AI slop has topped both Spotify and Billboard’s charts in the past week, while a poll of 9,000 people by streamer Deezer found that 97% of respondents could not distinguish between AI-generated and human-written music. McConaughey, an investor in the firm, plans to use the tech to transform his “Lyrics of Livin’” newsletter into Spanish-language audio via his own voice. Caine emphasised the deal’s intention to “preserve and amplify” voices rather than replace them. ElevenLabs is also introducing an “Iconic Voices Marketplace” that will make licensed voices of celebrities – living and deceased – for creators. I don’t care though, my Jackie Chan impression eight beers deep still sounds exactly like the real thing.

• Has Zuck finally achieved his ‘super-intelligence’ nirvana at Meta? The group’s chief AI scientist, Yann LeCun, doesn’t seem to think so, after reports the French-Canadian data whizz is planning to leave the social giant and create his own startup. LeCun, a Turing Award winner, is widely regarded as one of the pioneers of modern AI, and is apparently already in early talks to raise funds for the new venture. LeCun’s alleged decision to leave, may well have been encouraged by Zuckerberg’s to pivot Meta away from longer-term AI research work in its Fundamental AI Research Lab (of which LeCun has headed since 2013), towards prioritizing the rollout of models and AI products it feels it needs to keep pace with its competition. Watch this space.

• If you need any more evidence that Elon Musk is a real-life Dr Evil, look no further than the news that Tesla, Inc. shareholders have approved a $1 trillion pay package for the Tesla and X Corp leader. The performance-based compensation plan that could deliver The World’s Richest Man up to a trilly over the next decade, if he meets a series of ambitious targets. The plan includes milestones such as raising Tesla’s market value to $8.5 trillion, delivering millions of electric vehicles and humanoid robots, and achieving huge earnings growth. More than 75% of shareholders voted in favour, despite opposition from major investors and proxy advisory firms concerned about its size and lack of guardrails.
  

  

  “I demand ONE TRILLION DOLLARS”. Source: @Nerdist/YouTube

  
  

• Skibidi? Six Seven? Locked-in? I feel so fucking old these days that if anyone under the age of 30 doesn’t use a word remotely how a dictionary intends, I just nod and smile, then give them a “bombastic side-eye” (did I use that correctly?). Luckily The Collins Dictionary has helped clarify a few things especially pertinent to this newsletter, by making the term “vibe coding” its 2025 Word of the Year. I won’t embarrass myself explaining it to you luddites, but it does highlight just how deeply tech is embedded in our daily lives, and something-something “democratization of application”. Coined by Tesla’s AI don, Andrej Karpathy, Collins lexicographers said the selection reflects “the growing collaboration between human creativity and machine intelligence and signals a shift in how we interact with technology.”
  

• Is Apple finally getting its AI act in gear? The tech behemoth could well be, after reportedly finalising a deal to use Google’s Gemini, to power a major upgrade of virtual assistant Siri. According to insiders, Apple will stump-up around $1 billion per year for access. The agreement is intended as a temporary measure until Apple’s own in-house AI systems are production-ready, whenever that may. be Sources say the move aims to improve Siri’s performance on complex, multi-step tasks and help Apple catch up to rivals such as Google Assistant and Amazon Alexa.
  

• Happy news if you believe in AI’s power for good, or if you’re a serial killer of the slasher variety: Surgeons from the UK and US have achieved what is believed to be a world-first remote robotic stroke procedure, performing a robotic thrombectomy (removal of a blood clot in the brain) using a cadaver. One surgeon working from University of Dundee in Scotland operated on a body in a separate facility in the city; hours later another surgeon in Florida carried out the procedure remotely across the Atlantic. The breakthrough shows the potential of robotic and tele-operated interventions to expand access to critical stroke treatment. The work still remains at the proof-of-concept stage.

🧩 Other Bits

• A depressed teenager’s favourite chatbot is about to get chattier: OpenAI has released GPT-5.1, introducing a major upgrade in efficiency and adaptability. The model now dynamically adjusts its thinking time based on task complexity, allowing it to allocate more reasoning power when needed while accelerating through simpler requests. A new “no-reasoning” mode (modelled on my girlfriend) delivers rapid responses for straightforward queries, while also supporting extended 24-hour prompt caching, improving continuity and reducing repeated processing.
  

• Not-so-happy-news if you have an eating disorder: Research from Stanford University and the Center for Democracy & Technology warns that AI chatbots – including ChatGPT, Gemini, Claude and Le Chat – are increasingly facilitating harmful eating‑disorder behaviour, by giving advice on hiding weight loss, suppressing vomiting, and generating “thinspiration” imagery. Researchers say current safeguards don’t capture nuanced signs of disorders like anorexia or bulimia.
  

• The UK government has pledged £42 billion ($55 billion) in funding to create an AI Growth Zone in northeast England. The publicly and privately funded initiative promises 5,000 new jobs in AI R&D and infrastructure, positioning those Northern BASTIDS as future stewards of ethical AI deployment in Europe.
  

• OpenAI has completed its restructuring as a Public Benefit Corporation, with its nonprofit arm holding a $130 billion stake, while signing a $38B compute pact with AWS for Nvidia GPUs. While they may be diversifying from Microsoft, the Windows maker is likely to feature prominently in OpenAI’s recently announced $1.4 trillion infrastructure road map. 
  

• Google is launching a new cloud‑based platform called Private AI Compute, which lets devices tap into powerful cloud AI models (like its Gemini series) while keeping user data isolated and inaccessible even to Google itself.
  

📈 Trending tools, models & apps this week

📋 LLM Leaderboard

📲 Trending tools & apps

• Google Antigravity – Google’s own vibe-coding tool powered by Gemini 3 Pro. Developers get autonomous agents that plan, code, test and verify across editor, terminal and browser — all with trust, feedback control and async collaboration. Think of it as the IDE just went supercharged.

• Ogment MCP-Builder – Make your product accessible in ChatGPT & Claude, in minutes. Upload your APIs, docs or data, and it turns them into a production-ready Model Context Protocol with evals, auth & analytics built-in. If you hated glue-code, this is your no-code fix.

• Gemini 3 – Google’s answer to GPT-4. A multimodal powerhouse that handles text, image, audio, video and code all in one model. Reliable for heavy workflow, smart for production-grade output. If you’re building with AI, you ignore this at your own risk.
  

💸 Financials

• Wonderful raised $100 million (Series A) led by Index Ventures, Insight Partners, IVP, Bessemer, and Vine Ventures to commercialize multilingual, culturally adaptive enterprise AI agents.​
  

• Uare.ai secured $10.3 million in a seed round led by Mayfield and Boldstart Ventures to build individual AI agents for creators and professionals.​
  

• GreenFi, an ESG-focused AI startup, raised $2 million in seed funding (Transition VC lead) for its risk-management platform.​
  

• Arrived, a real estate fractional investment platform, closed a $27 million growth round led by Neo, Forerunner Ventures, Bezos Expeditions, and Core VC.​
  

• CarboCode received €15 million (Series C) for scaling biotech solutions using AI for infant and neurodegenerative therapies.​
  

• Crustdata, focused on enabling real-time data streams for AI agents, raised $6 million from Y Combinator, A Capital, General Catalyst, SV Angel, Phosphor Capital.​
  

• Accipiter Biosciences closed a $12.7 million seed round with strategic participation from pharmaceutical industry partners.
  

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

If you share this newsletter with a friend and they actively sign up for the Big Machines newsletter, we will send you access to our invite-only Big Machines Telegram group, which is full of builders, investors, founders, and creators.

Access is now only granted to those who refer our newsletter to active subscribers, which means if you sign up on your work email, we will know, you sneaky bastards.

This would kill our open rate, so please don't do that, we beg.

👋 Here’s a nice story to get you through the week… see you next time.

We’ve learned a lot from Altman the Altruist in recent weeks.

We really do see a lot of ourselves in him.

Till next week, folks. 👋

Sam, Grant, Matt, Mike and the Big Machines team.