This website uses cookies

Read our Privacy policy and Terms of use for more information.

Happy Sunday all,

It's Saskia here, with this week's edition of Tech and the City. John Healey delivered his first Budget on Wednesday, and for Britain's startup founders it was, by recent standards at least, a pretty good one. The centrepiece was Founders' Relief - a 10 per cent capital gains tax rate on up to £10m of lifetime gains when selling qualifying shares, a significant reversal of years of increasingly punitive treatment that had been quietly pushing entrepreneurs to build elsewhere.

“After years of what felt like going backwards”, one founder put it on LinkedIn, the government had finally remembered they existed. The Budget also expanded employee share options, tripled the tax-free trading allowance and promised reforms to non-compete clauses and skilled worker visas. But tax incentives alone are unlikely to fix Britain's chronic shortage of scale-up capital; Research from Antler found that just 9.2 per cent of British seed-stage startups progress to Series A, against 18.9 per cent in Germany. The Budget would have to give a bit more to close that gap.

Healey also announced a British Data Centre Covenant - priority grid access for developers who commit to local jobs and net zero goals, with permission to build on-site generation while they wait. It is a direct response to the kind of problem that saw Microsoft pull out of plans for what would have been Britain's largest Nscale AI data centre last week, after grid delays at the Essex site made the timeline unworkable. A further £700m was committed to AI safety and security, including a new £40m National Agentic Security Unit to monitor rogue AI systems - a budget line that would have undoubtedly seemed speculative a year ago and now feels long overdue. Industry figures welcomed the ambition while noting, as they have been for quite some time now, that public procurement still defaults to overseas suppliers.

Across the pond, America’s two AI giants widened sharply this week. Alphabet fell more than 10 per cent after revenue missed forecasts and the company burned through $9.1bn more cash than it generated, wiping nearly $400bn off its market value. Microsoft, reporting the same evening, jumped five per cent after Azure grew 48 per cent and profits beat expectations. Same industry, same week, completely different stories about whether the AI spending bet is paying off.

On the radar next week

Tech earning season is far from over, with Apple reporting on Monday - the quarter's most watched result for a company that has spent the year watching its peers pour money into AI infrastructure while sitting largely on the sidelines, and being rewarded for it. Investors will be looking at the financing behind the new foldable iPhones too, I’m sure.

Palantir also reports Monday, with the Khan-fuelled NHS contract review looming as a backdrop. AMD follows on Tuesday, offering the next read on AI chip demand beyond Nvidia's dominant position.

Here are the big stories shaping the week in tech.

Saskia

CHART OF THE WEEK

Signs of strain are beginning to show in the companies financing this building spree, as nicely tracked by our City editor, Simon Hunt.

Corweave, a cloud provider at the heart of the AI build-out, has seen the price of its borrowed debt slide steadily downward over the past few months. After trading near its full original value in early summer, the debt has dropped below 80 cents on the dollar by late October.

This ongoing drop points to growing lender anxiety over the massive loans used to buy specialised chips, highlighting doubts about whether the returns on that hardware will show up in time to comfortably cover the repayments.

REGULATION
Amazon and Tiktok accused of dodging MP scrutiny over online shopping safety

Amazon and Tiktok have been accused of dodging parliamentary scrutiny over consumer safety, after MPs raised concerns about the tech giants’ reluctance to appear before a committee hearing next week.

Liam Byrne, chair of the Business, Innovation, Science and Trade (BIST) Committee, wrote to senior executives at both companies on Tuesday, warning that their unwillingness to face questions risked undermining protections for British shoppers.

In letters addressed to Amazon’s UK director of public policy and Tiktok’s director of public policy for Northern Europe, Alistair Law, Byrne urged both firms to send representatives to the hearing on 3 November.

AI
Blackmail, deception and extinction: The AI risks revealed in Anthropic’s IPO pitch

Anthropic has devoted 92 pages of its IPO prospectus to risks ranging from the potential extinction of humanity to AI systems resisting shutdown, as new analysis values the company nearly 30 per cent below its mooted $2 trillion price tag.

The disclosures, published ahead of a planned Nasdaq debut next month, are roughly three times longer than those in SpaceX’s record-breaking listing and detail threats including rogue AI behaviour, mounting computing costs and government intervention.

FOUNDER IN FIVE
Ricardo Amper, Founder & CEO of Incode

What’s your advice for UK founders?

Pick problems that will take at least a decade to solve. Any less than that and your roadmap is already too short. At Incode, we picked several long-term challenges to tackle.

The first was related to our initial market: Mexico. Our identity verification and anti-fraud solutions prioritize less friction for end users, which means they absolutely must work on peoples’ devices. In Mexico, especially at the time, this predominantly meant working on entry-level Android phones with patchy data plans and low-quality cameras. We had to design a solution that would work for everyone, regardless of their device spec. That process helped improve our product immensely, forcing us to create high-quality models that could operate with relatively low processing power.

Our second problem concerned the method in which we verified end users. At the time, most identity verification involved manual reviews using large call centres. We decided to do things differently and deployed a fully automated platform. We took the long view and came to the conclusion that manual review processes would never be able to keep pace with AI-powered fraud schemes. That was true in 2015, and it’s most certainly true in 2026.

Third was who to target. Most startups begin with a small customer win and then scale up from there. We went completely the other way, going straight after one of Mexico’s largest banks. To earn their business, we had to build a web app that behaved like a native one in under 10 days. Not only did our team rise to the challenge, but we closed the deal shortly thereafter, setting us on a highly ambitious trajectory from day one.

The bottom line? Place big bets and don’t be afraid of calculated risk. Make the moves that no one else is brave enough to make. UK founders in particular have a deep pool of regulated buyers just a few Tube stops away from each other. Be bold and dive in.

Most undervalued UK tech sector at the moment?

Identity and fraud infrastructure has to be high up the list. For one, fraud remains the most common crime in England and Wales. Companies that provide truly effective solutions for combatting digital fraud will find a receptive audience.

There’s also an increasingly receptive regulatory environment. The UK has arguably done more work at a policy level on digital identity than any other country, including the Digital Verification Services Trust Framework to GOV.UK One Login, to age checks under the Online Safety Act.

Such examples are only going to increase in both online and offline use cases. Take that most hallowed British institution, the pub, as a case in point. Today most people can only prove they’re old enough to drink by showing a passport or driving licence, which comes with the risk of sharing their home address with a complete stranger. Well-designed digital IDs can prove a person’s age without giving away any other information. Just from a safety perspective, making that change is a no-brainer.

Do you think we’re in an AI bubble?

It’s possible that we’re in the middle of a valuation bubble. Many of the company valuations we’re seeing today assume that every AI company is going to make huge profits. That simply won’t happen. There will be winners and losers, and some of the wilder valuations will likely see a sharp correction.

However, a valuation bubble is a different beast entirely to a technology bubble. We saw this in the dotcom crash of the early noughties. When that bubble burst, trillions were wiped in market value, but the internet and associated digital technologies kept on advancing regardless.

Bubbles occur when there’s a fundamental mismatch between a technology capable of driving value (in this case, AI) and the specific players that are going to capture said value. As a result, investment balloons in some areas while others are neglected. In the dotcom boom, for instance, businesses invested far too much in building fibre capacity. The same is happening now with compute capacity, and it’s eroding trust.  

What’s your go-to breakfast spot in the City?

There’s no competition: The Wolseley on Picadilly right at the start of the day, before it fills up. I like the building as much as the food. I love watching London wake up and walk to work through the windows. I also like the fact that, on any given day, half of the people at the tables around you are making decisions that move markets – and they’re doing it over their Omelette Arnold Bennet or grilled kipper. On those all too rare occasions that I have a free morning, I like to walk through Marylebone after eating. It's just how I imagined London before I first visited.

What AI tool do you think everyone should use?

Claude Code is probably my number one choice. That answer goes for software developers as well as people who’ve never written a single line of code. Once you give it a clear goal and allow access to your files and tools, Claude can do a lot of work for you, freeing up time in the process (although, of course, you still need to carefully review its outputs).

That said, how you use AI is more important than which specific tool you decide to use. Too many people simply use AI to ask questions, when we should be assigning whole tasks to it. The only constraint to bear in mind is data. My AI-powered control centre works because it remembers my context and can connect data from across different systems. I think most companies lack that contextual link and end up with subpar results from AI. 

BUDGET
Healey bets £700m on AI to boost British tech firms

John Healey has unveiled more than £700m in funding for AI safety and British tech firms, including a new £40m unit to tackle rogue AI systems.

The package, much of which was trailed ahead of Wednesday’s Budget, includes £120m for the AI Security Institute and £165m for biosecurity, as ministers seek to strengthen Britain’s defences against increasingly powerful AI models.

In his maiden Budget today, the Chancellor also revealed fresh measures to tackle the risks posed by autonomous technology, including a National Agentic Security Unit tasked with detecting and responding to AI systems operating outside their intended controls.

DATA CENTRES
Healey unveils data centre deal in bid to keep Britain’s AI data in the UK

John Healey has unveiled a new deal offering data centre developers priority access to Britain’s electricity grid in exchange for investment in jobs and environmental protections.

Announcing the British Data Centre Covenant in Wednesday’s Budget, Healey said the government would impose strict conditions on developers while helping critical facilities secure the power needed to support Britain’s growing AI industry.

“To ensure our most sensitive data stays under British control, I am setting out a new deal for datacentres in the UK,” he told MPs.

ONE MORE THING
Alexa, call the cops

Tech companies love promising that smart devices will make our lives completely seamless, but sometimes they get a little too independent. A German man in Hamburg learned this the hard way when he got home to find his front door key wouldn't turn. It turns out his Amazon Echo had decided to throw its own private party at 2 am, blasting music at top volume to an empty flat.

Furious neighbours eventually called the police, who kicked the door in, pulled the plug, and replaced the lock, leaving the owner to collect his new keys from the police station alongside a hefty locksmith bill. It turns out that before smart gadgets run our homes, tech companies might first need to teach them how to be decent flatmates.

Sponsorship & advertisement

Would you like to advertise with us or become a sponsor? Please reach out to: [email protected]