Revolut soars

Author

Alex Wilhelm
November 24, 2025 • Estimated Reading Time: 8 minutes

Top News

  • Revolut now worth $75B: Following a lengthy fundraising process, British neobank Revolut is now valued at $75 billion, thanks to new capital from Coatue, Greenoaks, and other investors. With plans to enter thirty new markets, Revolut aims to continue growing at a rapid pace. In its 2024 report, the digital bank reported 72% revenue growth to £3.1 billion, along with a profit before tax of £1.1 billion, up from just £438 million in the previous year. Naturally, that data is somewhat dated at this point, but investor interest in adding $30 billion to Revolute’s valuation implies that it had a pretty good 2025, too.

  • AWS bets on government AI: Amazon’s AWS cloud computing group plans to invest up to $50 billion in building out government-focused AI capabilities. Starting in 2026, AWS will “add nearly 1.3 gigawatts of AI and supercomputing capacity across AWS Top Secret, AWS Secret, and AWS GovCloud” by adding more data centers to their lineup. The bet here is that the government’s current push to utilize more AI will persist into the future, regardless of which political party is in charge. Cue the same from Azure and Google Cloud?£

  • Boat-based fusion power: Humans have scant history of producing power-positive fusion reactions, but that isn’t stopping startup Maritime Fusion from hustling to put the futuristic power source on boats. Maritime thinks that High Temperature Superconducting (HTS) magnets won’t do well as grid-level power sources, but will do just fine on ships, allowing it to replace today’s carbon-intensive at-sea power generation techniques with something far cleaner. Naturally, the startup faces technological challenges, but who doesn’t love a big vision for an even bigger market? (HTS magnets are having a good month, too.)

TWiST 500

The TWiST500 features a host of companies building AI models, some of which are based in China. Today, the list includes DeepSeek and Moonshot.ai. Moonshot, in particular, is on a hot streak thanks to its performant Kimi K2 Thinking model. But as we wrap the TWIST500 itself, we’re adding Minimax and Z.AI. MiniMax’s M1 (large) and M2 (small) models have posted incredible benchmark results, while Z.AI’s GLM-4.6 model compares favorably to models from Anthropic and elsewhere.

Even better, Chinese AI labs are pursuing an open-source approach, offering their technology with open weights and often more. For startups and developers, having access to open-source models is a boon as it forces closed-source AI labs to invest in improving their own technology even faster; it’s hard to compete with free, after all.

But why? Why are so many Chinese AI labs taking an open approach to building models while the American rivals are nearly entirely closed-source? A conversation with Zixuan Li, Z.AI’s director of product, explains:

[A]s a Chinese company, we need to really be open to get accepted by some companies because people will not use your API to try your models. Maybe they deploy on Fireworks, maybe they use it on Groq, or maybe they download to their own chips. It’s not easy to get famous in the United States because people just don’t accept your API. They need to be stored in the US. I think it’s necessary to be open right now for people to use our models.

That’s plainly stated and true. The good news for Z.AI and its peers is that there’s a somewhat proven business model for the open-source approach to building AI models, as Zixuan pointed out:

We also learned from DeepSeek because our flagship model was closed source back in 2024. But when DeepSeek R1 launched, we realized that you can be really famous for open sourcing your model while getting some business return through API or other collaborations. You need to expand the cake first and then take a bite of it.

I love the smell of competition in the morning. — Alex

A message from Uber AI Solutions

Uber AI Solutions: Your trusted partner to get AI to work in the real world. Book a demo with them TODAY at Uber.com/twist

This Week in Startups

E2213: Google and Meta had their cases dismissed (or received a slap on the wrist)… Despite all the backlash and cynicism, AI companies continue making bank and releasing hot new products… What does it all mean? For Jason Calacanis, the signs are pointing to a “major M&A moment,” with huge opportunities for increased efficiency and consolidation among America’s favorite brands and largest companies?

E2212: Mistral CEO Arthur Mensch stops by the show to tell us how enterprises can start getting more out of their AI pilots, why the West seems to lack leadership around open-source AI, and teach Jason and Alex the science of “benchmaxxing” (that is, focusing on your AI model scoring better on tasks rather than actually getting smarter). PLUS why everyone made the wrong call on OpenAI vs. Google, Kraken’s $20 billion valuation, Jason’s urgent M&A message, and more.

E2211: Grok 4.1 and Gemini 3 went HEAD TO HEAD this week, launching right on top of one another. But how do we really KNOW which who’s leading the seemingly endless AI Death Race? On a new TWiST, Jason and Alex survey the current AI landscape, and try to look beyond the mega-deals and benchmarking stats to get a real sense for the state of play. Plus Cloudflare shuts down the internet, how to turn off “Learn From This App” on iOS, ChatGPT takes a nosedive on Polymarket’s “Top AI” market, and more.

TWiST Partner Offers

  • Squarespace: Turn your idea into a beautiful website! Go to http://www.squarespace.com/twist for a free trial. When you’re ready to launch, use offer code TWIST to save 10% off your first purchase of a website or domain.

  • Sentry: Your team should be focused on shipping features — not chasing down bugs. New users get 3 months free of the Business plan which covers 150k errors. Go to sentry.io/twist and use the code TWIST

  • Enterpret: Enterpret turns feedback noise into Customer Intelligence, so your team knows exactly what to fix and build next. Head to Enterpret.com/twist to book a demo and see it in action.

Founder University is headed to Tokyo!

Applications are now open for the 2026 cohort. Selected startups will take part in a global, three-part experience designed to accelerate growth and expand international reach:

  • Tokyo Bootcamp: January 12–16

  • Online Acceleration Program: January 19–March 6

  • U.S. Travel Program: March 9–20

If you’re building something big and ready to go global, this is your moment. Apply now at launch.co/tokyo.

Investor Networking Lunch in SF

Jason and LAUNCH are hosting an investor networking lunch, followed by pitches from our latest Accelerator batch. We'd love for you to join us for Dim Sum in San Francisco on Friday, December 5th. (Or forward this to your investor friends.)

  • Date/Time: Friday, December 5th, 2025 at 12-3PM PT

  • Format: Dim sum lunch, followed by founder pitches & investor networking

  • Location: TBA (Chinatown area)

Spots are limited, please RSVP here. The LAUNCH team will email you confirmation once your registration is accepted. You must be an accredited investor or work for a VC Fund to join.

SF Live-Work Space Now Available

Need a flexible living and working environment in San Francisco? This thoughtfully designed loft-style residence at 787 Bryant St., the heart of the vibrant SOMA district and the city’s creative hub, is now available for rent or purchase. Check the listing for more details.

The TWiST500 newsletter is the new, updated, and improved TWiST Ticker.