A bitty but important week! 55% of you said to keep links at the bottom so they’ll stay there for the time being, but I may play with the layout…

EARLY BIRD OFFER ALERT! Finally moved all WD_DTW titles inside Beehiiv, and finalised the launch strategy for ‘What Did Meta Do This Week?’, you’re the first to know it will launch in August, and there’s an early bird offer of £4.95 a month a year for the first 250 people so get in quick.

Also completed the decks for two august speaking opps (superintelligence disruption + where FMCG brands are leaving money on the table). Also landed on the tool for the subscription paywall for POTATO, so if you’re in UK tech, get in there at the £5 per month level before it doubles. Also worked on BILL[AI]ABLE schedule and ecosystem - exciting stuff if you’re in the agency world. Like I said, bitty but important!! Subscription for the win!

Finally, a big thanks for all the LinkedIn posts praising TNN, and I am glad you love the new space and our new blend of coffee from Apostle! The next TNN is an important topic: leadership and personal performance with Sifu Julian Hitch. Mark your calendar for August 21st at Oneder. This one is a must attend with what’s coming over the next 24 months…

Meta detailed its superintelligence roadmap and defended political ad policies in the EU, while its new chief scientist came from (you guessed it) OpenAI. Meta, Apple, and Google clashed over age verification laws, with facial recognition now required to access adult content in the UK via selfie (of course there’s all kinds of work arounds). Apple launched AppleCare One for $20/month and faced deeper UK antitrust scrutiny. Apple also broadened app age ratings, citing child protection and definitely not to fend off regulatory pressure and shift liability to developers.

YouTube posted strong Q2 earnings, and expanded Shorts with AI image-to-video tools. Gemini now powers 2B monthly AI Overviews and is testing virtual try-ons and web view search experiments. Microsoft confirmed SharePoint hackers are deploying ransomware, and the US government probed the SharePoint breach’s national security impact. OpenAI was outperformed on reasoning benchmarks by Alibaba’s open-source Qwen 3, while its fundraising drew $1B+ from Founders Fund and Dragoneer.

Amazon acquired Bee AI, launched a new Kindle ColorSoft for kids, and faced criticism over AI agents planting computer-wiping code. Nvidia avoided a breakup threat from Trump and was cited in China’s repair demand boom. Huawei showed off a direct competitor, deepening compute rivalries. IBM posted lukewarm software growth, Intel disappointed with earnings, and AMD flagged 5–20% higher US chip costs. Uber tested a female-driver option in the US, Lyft plans to launch autonomous shuttles, and Sonos made Tom Conrad CEO.

▲ The (terrifyingly exciting/terrifying) economics of superintelligence. /8 mins

Roofball is a real thing, and blowing up (apparently). /5 mins [I dare you to spend more]

Germany’s Wendelstein 7-X fusion device held superhot plasma stable for over eight minutes, setting a record and showing that fusion reactors could one day run continuously to produce clean energy. /5 mins

DuckDuckGo now lets you hide AI-created images in search. /3 mins

How to run an LLM on your laptop. /10 mins

Before GLP-1 inhibitors storm the world, there may be a pill that burns fat without suppressing appetite. /8 mins

Almost half of working-age adults in the UK are not putting any money into a private pension at all. /7 mins

LinkedIn execs are saying AI is killing the career ladder. /6 mins

The world wide web of everything: how spatial technologies are about to rewire the real world. /7 mins

Quaise, the geothermal energy startup, wants to use beams of energy to drill geothermal wells. /12 mins

Are you ready for vibe analytics? /5 mins

Meta’s newly launched Superintelligence Lab marks the company’s most aggressive pivot in a decade, signalling not just a technical escalation but a foundational redefinition of what Meta intends to be. Over the past month, Zuckerberg has orchestrated a hiring spree that few outside of OpenAI or DeepMind could rival—recruiting Shengjia Zhao (co-creator of ChatGPT), Daniel Gross (former CEO of Safe Superintelligence), Jason Wei (co-author of OpenAI’s o3), and multiple senior AI researchers from Apple, Waymo, and beyond. The money is blunt: $200 million compensation packages, poaching entire teams, 49% of Scale AI at a ~$30 billion valuation. The ambition is unsubtle: to lead, not just compete, in the race for superintelligent AI.

The first 250 people to sign up to 'What Did Meta Do This Week?' will get it for £4.95 a month for the first year when it launches in August.

Meta is one of the most consequential companies in the world. With over 3 billion monthly users across Facebook, WhatsApp, Instagram, Threads, Messenger, and Reality Labs, its reach is unparalleled. Zuck and co are reshaping competitive AI strategy across the tech sector, while facing unprecedented internal and external pressures. No firm is simultaneously as exposed, aggressive, or essential to track.

For years, Meta operated in the open-source lane—prioritising distributed research, publishable papers, and interoperable tooling like LLaMA. That posture may be changing. According to internal reports, Meta is now considering abandoning Behemoth, its flagship open-weight model, in favour of a closed model architecture within the new Superintelligence Lab. The team, led by Alexandr Wang (of Scale AI fame), is stacked with top-tier theorists and infrastructure builders. This is no longer about catching up to OpenAI, it’s about leapfrogging the field with a vertically integrated, closed-loop intelligence stack.

Zuckerberg is not hiding the stakes. In recent remarks, he said building superintelligence is “the most important problem of our generation” and confirmed that Meta will pursue it directly and unapologetically. Framing like that matters to more than just column inches. While Apple and Google hedge behind personal assistants and product integrations, Meta is declaring that the endgame is AGI. The name of the lab is not metaphorical.

The move reshapes Meta's internal strategy and reconfigures the competitive map. OpenAI still holds a technical edge, but Meta now has the cash flow, infrastructure, and momentum to build at the frontier. Google is stretched thin between Gemini's multiple product lines. Anthropic is deep on safety and interpretability but doesn’t have Meta’s scale. Apple remains focused on assistant-level AI. If Meta succeeds in fusing raw compute with bleeding-edge reasoning research, it could become the dominant closed-loop intelligence company in the West. The cost, however, is real in a lot of senses.

Abandoning open-source AI weakens Meta’s developer goodwill. Copycat accusations will mount, especially as the lab grows more secretive. Regulatory scrutiny is already intensifying as Meta attempts to shift from platform owner to intelligence provider. There are also ethical risks. Superintelligence work demands unprecedented transparency, and Meta’s track record on trust is brittle at best. Will governments, developers, or users believe Meta can safely steward post-human-level intelligence? It’s never been more critical to know which way Zuck and co want the wind to blow.

SO WHAT?

Meta is no longer playing the same AI game as everyone else. The Superintelligence Lab is a moonshot with real teeth, backed by revenue, talent, and ruthless focus. Recents moves are the clearest indicator yet that Zuckerberg is done chasing product-market fit and now wants to shape the market itself. The stakes are existential. If Meta builds a controllable, high-performance AGI before anyone else, it will control the next interface, the next economy, and potentially the next definition of intelligence. If Meta doesn’t do this they’ll slowly be edged out and become irrelevant.

For the industry, this signals consolidation. Smaller labs may struggle to retain talent or stay independent. Governments will face rising pressure to regulate foundational models at a geopolitical level. OpenAI, long seen as the incumbent, is suddenly flanked by a company with equal talent and far more distribution. The winners are Meta’s infrastructure partners, its top researchers, and any developer able to plug into the superintelligence stack early. The losers may be transparency, alignment research, and the open AI movement more broadly.

Zuckerberg has made his move. The rest of the ecosystem now has to decide whether to follow, compete, or more likely, totally resist.

[DO] Sign up to ‘What Did Meta Do This Week?’, nowhere else covers everything Meta does in one place.

[DON’T] Make sure you get the best deal on ‘What Did Meta Do This Week?’. That’s either to £5 a month or get all the titles (OpenAI, Amazon, Google) for £49.99 along with a membership to TBD. Check out the Explorer tier here - it saves you £400 a year (and more when the prices increase).

Advisor to global brands and agencies, he translates frontier tech into strategy that protects relevance and unlocks growth. Founder of TBD Group, author of Disruptive Technologies, and a go-to analyst for FT, WSJ, BBC, and CNN.

Need a second opinion on what’s next?
Find out more and connect with him on LinkedIn and Bluesky.