Are VPNs Banned in the UK? Jensen Huang on AI Doom

Jensen Huang says 2030 is not the end of the world. We break down the AI doomsday myths, real costs, and what to actually do. Plus: are VPNs banned in the UK?

## Jensen Huang Says 2030 Will Not Be the End of the World. Here Is Why That Matters for Your Wallet Are VPNs banned in the UK? No, they are not, and that question is worth answering up front because it tells you something about how fast AI doom predictions age. In 2023, people were convinced the UK would outlaw encrypted tunnels within a year. It did not happen. The Online Safety Act got messy, sure, and the Investigatory Powers Tribunal still argues with Apple, but you can still buy a Mullvad subscription in Manchester this morning. Doom predictions have a terrible track record. Nvidia's Jensen Huang made the same point about AI this month. His line, roughly, was that 2030 is not going to be the end of the world. Coming from the man who sells the shovels in this gold rush, you might dismiss it as self-interest. You would be half right. Huang has every incentive to keep the optimism tap running. But he also has better visibility into actual AI deployments than almost anyone alive, and his argument is more specific than the usual "AI is good, trust me" routine. So should you panic about Jensen Huang AI doomsday talk? No. Here is what is actually happening, what it costs, and what a sane person does about it. ## The Doomsday Case, Steelmanned Let us not strawman the pessimists. Serious people worry about three things. 1. **Job displacement at scale.** Not "AI writes your emails" but entire categories of white-collar work compressed into a few years. Entry-level legal review, junior copywriting, first-line customer support. 2. **Concentration of power.** Four or five companies control the compute, the models, and the distribution. That is not a healthy market. 3. **Alignment risk.** The tail scenario where a sufficiently capable system does something catastrophic. Low probability, enormous downside. None of these are crazy. The problem is the timeline. Doom discourse tends to compress 20 years of messy, uneven, regulation-buffeted change into a single dramatic Tuesday. That is not how technology diffuses. Electricity took 40 years to reorganize factories. The internet took 25 to eat retail. AI is moving faster, but it is still moving through institutions, procurement cycles, and union negotiations, all of which are slow. ## What Huang Actually Argues Huang's position, stripped of keynote polish, comes down to three claims. ### Compute demand is real, not speculative Nvidia's data center revenue is not a bubble in the dot-com sense. The chips are being bought by companies with actual revenue attached to the workloads: cloud providers, sovereign AI programs, banks running fraud detection, pharma companies doing protein folding. When your customers are Microsoft, Saudi Aramco, and Novartis, the demand curve has a floor. ### The scary capabilities are further out than the discourse suggests Huang consistently points out that today's models are narrow specialists. They are extraordinary at pattern completion and terrible at long-horizon planning. The gap between "writes convincing text" and "autonomously runs a company" is not a small increment. It is a different class of problem, and nobody has a credible path to it yet. ### Regulation will shape deployment, not stop it The EU AI Act is now in phased enforcement. The UK has its own lighter-touch regime. The US is a patchwork. None of this halts the technology. It slows specific applications in specific jurisdictions, which is exactly what regulation is supposed to do. ## A Worked Example With Real Numbers Say you run a 12-person digital agency in Berlin. Your biggest cost is junior staff doing research, first-draft copy, and client reporting. Total payroll for those roles: about $420,000 a year. You adopt AI tooling in 2026. Here is a plausible twelve-month picture. - **Tools:** [ChatGPT](https://chat.openai.com/) Team, Claude Team, Cursor, plus a [Perplexity](https://www.perplexity.ai/) Enterprise seat. Roughly $180 per person per month across the team, so about $26,000 a year. - **Productivity gain:** Juniors now produce first drafts in 20 minutes instead of 3 hours. You do not fire anyone. You take on 40 percent more clients with the same headcount. - **New revenue:** At an average retainer of $4,500 a month, three extra clients is $162,000 a year. - **Net effect:** $136,000 improvement, no layoffs, and your juniors now have a skill (AI-augmented workflow) that makes them more employable, not less. That is the boring, real version of AI's impact. Not utopia, not apocalypse. Margin expansion and role redefinition. Now the pessimistic version. Same agency, but leadership decides to cut four juniors and pocket the savings. Six months later, senior staff are drowning in the work juniors used to absorb, quality drops, two clients leave, and the firm is smaller than when it started. The technology did not cause that. Management did. ## Our Take: What to Actually Do Panic is not a strategy, and neither is pretending nothing is changing. Here is what we would tell a friend over coffee. **Buy the tools, keep the people.** The agencies and teams winning right now are the ones using AI to expand output, not shrink headcount. Cutting juniors looks efficient for two quarters and then bites hard. **Learn the workflow, not the hype.** You do not need to understand transformer architecture. You need to know which tasks AI does well (summarization, first drafts, code scaffolding, data cleanup) and which it does badly (judgment calls, relationship management, anything requiring accountability). **Ignore the 2030 countdown.** Anyone selling you a specific doomsday date is selling something. The realistic timeline is a decade of uneven change, with plenty of room to adapt. **Watch the regulation, not the rhetoric.** The [EU AI Act](/tech/blog/ai-safety-showdown-what-the-doomsday-debate-means-for-everyday-users), UK policy, and US state-level rules will shape what you can deploy far more than any model release. If you are in a regulated sector, this is where your attention belongs. For hardware, if you are building locally rather than renting cloud, an RTX 5090 at around $2,000 is the practical entry point for serious inference work in 2026. Below that, you are better off with API access than a mediocre local setup. ## The VPN Question, Answered Properly Since it comes up constantly in our inbox: are VPNs banned in the UK? Still no. The UK has tightened rules around age verification and encrypted messaging in specific contexts, but commercial VPN use remains legal and unregulated. The same is true across the EU and the US. If you are using a VPN for privacy, work, or streaming, nothing has changed. If you are using one to evade a specific legal restriction, that is a different question with different answers, and you should talk to a lawyer rather than a tech blog. The reason this matters to an AI story is that the same pattern repeats. A scary headline appears, everyone assumes the worst, and eighteen months later the actual rule is narrower and more boring than anyone predicted. AI regulation is following the identical curve. ## FAQ ### Is Jensen Huang right that AI will not end the world by 2030? He is right that no credible evidence points to a 2030 catastrophe. He is also a biased source, since Nvidia profits from AI adoption. The honest position is that near-term disruption is real and manageable, while long-term tail risks deserve serious research funding, not panic. ### Should I change my job because of AI? Not because of AI alone. Change jobs if your specific role is mostly routine text generation, basic data entry, or first-line triage. Those are the tasks AI handles well today. Roles requiring judgment, accountability, or physical presence are safer than the headlines suggest. ### Are VPNs banned in the UK in 2026? No. Commercial VPN use is legal across the UK, EU, and US. Recent UK legislation targets specific harms like age verification and encrypted messaging in narrow contexts, not general VPN usage. You can keep your subscription.

Are VPNs Banned in the UK? Jensen Huang on AI Doom

Frequently asked questions

Related articles