Happy Sunday all,
It's Saskia here, with this week's edition of Tech and the City. I spoke to Geoffrey Hinton this week - the Nobel Prize-winning scientist whose research helped lay the foundations for modern AI - and he was not short of opinions. “I think it's particularly dangerous because of the character of some of the people involved, like Altman, Musk and Zuckerberg. I wouldn't buy a used car from any of them”, he said.
On the idea of a kill switch to cut off rogue AI systems, he was sceptical: “When AI is that smart, it'll be very good at persuasion. So even if you have someone who's meant to turn it off if it misbehaves, it'll be able to persuade them not to”. And on Britain's prospects of becoming a serious force in the global AI race, he was characteristically direct: "The UK sort of tries to position itself as very important in AI. I don't think it's going to be."
Ministers who have spent the past year pitching Britain as a leading AI destination, and who last week committed another £60m to the Sovereign AI Fund, may find that assessment inconvenient. It is not, unfortunately, an uncommon one.

Markets had their own verdict on the state of the industry this week, as a sharp tech rout wiped hundreds of billions off valuations across the sector. The sell-off, driven by renewed anxiety over whether AI spending is generating returns quickly enough, dragged down chip stocks and cloud providers alike - a reminder that the gap between capital deployed and revenue demonstrated remains one of the more uncomfortable facts in the AI investment thesis.
On the radar next week
The FTT Fintech Festival runs Monday and Tuesday in London a useful read on where the City's appetite for AI-driven financial services actually stands, now that the initial enthusiasm has collided with the reality of implementation. The FCA's growing scrutiny of agentic AI in financial services means the conversations there will carry rather more regulatory weight than they might have done a year ago.
The week's main financial event, however, arrives the following Monday: Nvidia reports on 17 November, and after the market turbulence of recent weeks, the numbers will be parsed with unusual intensity. The question is no longer whether demand for AI chips is real (it clearly is) but whether the economics of the buildout make sense at the pace it is currently being pursued. Jensen Huang's forward guidance will do more to set the mood for the final weeks of the year than almost anything else on the calendar.
Here are the big stories shaping the week in tech.
Saskia
Chart of the week

Anthropic is leaning heavily on Claude to engineer its own replacements. According to an updated IPO filing on 2 November, the AI lab is nearing an inflection point where artificial intelligence effectively doubles the pace of its research, compressing two years of typical progress into just twelve months.
The strategy, internally dubbed “closing the loop,” involves using Claude to write code, clean training sets, run evals, and stress-test newer architectures. Anthropic has poured nearly $28 billion into R&D since early 2024, and the internal payoff is scaling fast: between January and September 2026, the list-price value of the compute Anthropic spent using its own models jumped over 250x. While the lab is still verifying whether it has officially cleared this self-acceleration threshold under its Responsible Scaling Policy, the recursive loop of AI building better AI is clearly underway.
BT boss refuses to rule out job cuts after ‘unprecedented’ Talktalk rescue

BT chief executive Allison Kirkby has refused to rule out redundancies among 900 Talktalk employees joining the telecoms giant, defending its controversial takeover of the broadband provider as an “unprecedented” rescue.
The FTSE 100 boss told reporters she could not guarantee the jobs would be protected while the deal remained under regulatory review, as BT presses ahead with its own sweeping workforce reductions.
The company acquired Talktalk’s consumer and wholesale businesses out of administration last month, taking on services used by 2.5 million customers in a deal expected to cost around £400m this financial year.
Sequoia scraps London office as UK employees face axe
Silicon Valley venture capital giant Sequoia Capital is shutting its London office and cutting staff, just six years after expanding into Europe, in a major retreat from Britain’s startup investment scene.
The US investor, an early backer of Apple, Airbnb and Stripe, confirmed the closure in a sweeping restructuring announcement, saying it no longer saw the need for a physical presence in the capital.
“We are closing our London office, which means several people will be leaving Sequoia”, the firm said in a blog post published on Wednesday.
Government intervenes in Sky’s £1.6bn ITV takeover

Culture secretary Lisa Nandy has formally intervened in Sky’s £1.6bn takeover of ITV’s broadcasting business after raising concerns over its impact on media competition and the diversity of British news.
Nandy confirmed on Wednesday that she had issued a Public Interest Intervention Notice (PIIN), instructing the Competition and Markets Authority (CMA) and media regulator Ofcom to look into the proposed deal.
Both watchdogs have until 27 January to report their findings, after which Nandy will decide whether to order a more detailed investigation.
Cold war: UK food suppliers hit by European cyberattacks
British food suppliers have been caught up in a wave of cyberattacks targeting supermarket refrigerators across Europe, prompting warnings that outdated technology is leaving the UK’s food supply chain exposed.
The National Cyber Security Centre (NCSC) confirmed on Tuesday that vulnerable refrigeration technology had been exploited in a small number of incidents in Britain, after hackers remotely manipulated supermarket fridge temperatures across several European countries.
The attacks have forced retailers in Sweden, Belgium, Ireland and elsewhere to throw away large quantities of fresh and frozen food after refrigeration systems were switched into defrost mode.
Dr Farzana Rahman, Co-founder and CEO, Hexarad

What’s your advice for UK founders?
Don’t let rejection put you off. It’s part of the process, and everyone who has built something has a pile of “no”s behind them. Whether that’s doing your first investment round, trying to sign your first customer, or trying to hire your first key personnel, you’re going to hear a lot of “no” and that’s okay. Developing the confidence to overcome rejection and find the people who will believe in and support you is key.
Most undervalued UK tech sector at the moment?
Health tech! For all its challenges, the NHS is one of the most comprehensive healthcare systems in the world, and it’s free at the point of access. It has scale, data and a single system serving an entire population - all of which makes it a place ripe for innovation. There are so many problems waiting to be solved, from how patients move through the system to how clinicians spend their time. Too often at the moment, health tech is seen as fintech’s poor cousin. I’d love to see that change, because the potential impact on people’s lives is enormous.
Do you think we’re in an AI bubble?
AI is fundamentally changing how the world works, so I think its prominence in the conversation is appropriate. But invariable that also means that huge sums are being invested in areas that may never deliver much impact. There’s a real sense of urgency and fear of missing out, and that tends to lead to poor decision-making. But I don’t think that undermines the bigger picture. The companies that come through will be the ones solving genuine problems rather than simply riding the wave.
What’s your go-to breakfast spot in the City?
Dishoom in London Bridge. My co-founders and I have a weekly early breakfast meeting there. The refillable chai is a great way to start the day.
What AI tool do you think everyone should use?
At work it’s currently Claude. For my personal life admin, I’ve been using Fo by Wajo, which a good friend recommended to me. Between the two, they take a lot of the small, time-consuming tasks off my plate
Anthropic’s growth slows ahead of IPO as AI spending bill nears $900bn

Anthropic’s revenue has surged twelvefold to nearly $17bn (£13bn), but the AI giant has slipped back into the red as its computing commitments approach $900bn ahead of a potentially record-breaking stock market debut.
The Claude maker revealed the figures in an updated IPO filing on Monday, offering investors a fresh look at its finances just weeks before its expected Nasdaq listing.
Revenue jumped from $1.4bn a year earlier to almost $17bn in the three months to September, although quarterly growth slowed sharply to 47 per cent, compared with 144 per cent in the previous quarter.
One more thing: Oppenheimer never won the Nobel Prize in physics
A week after Geoffrey Hinton was announced as a Nobel Prize winner - in physics, a subject he does not work in - he was giving a lecture to a bank in New York. As he told me, he was making money “because I'm unemployed”. The final question from Tuesday’s event was whether, like Oppenheimer, he had come to regret what he had built. Hinton paused. “Something flashed through my mind”, he said, “and I simply couldn't resist it”. His answer: “Oppenheimer never won the Nobel Prize in physics”. Perhaps the most elegantly credentialled deflection in the history of existential risk.
Sponsorship & advertisement
Would you like to advertise with us or become a sponsor? Please reach out to: [email protected]
