Everyone wants a piece of the stablecoin boom

Also: A true shot for European tech companies to grow is here

Welcome to Cautious Optimism, a newsletter on tech, business, and power.

• 📈 Trending Up: AI models, after Google shipped Gemini 2.5 … which appears to be very solid at writing code … agents … not IRS agents … Chinese AI models for the Chinese market … Anthropic’s ability to use copyright-protected information to train, kinda … patronage … Qualcomm-ARM relations …

  • Chagee, a chain of Chinese consumer beverage shops, is going public here in the United States. It has some very impressive financial results to boast about. Still, after the Luckin Coffee mess from an era ago, I wonder how much thirst for its equities domestic investors will show.

  • Alex Konrad, a lovely human previously of Forbes, is building his own media empire. More here.

• 📉 Trending Down: Public broadcasting … our shared reality … open-source … layers of management … Nvidia chips in China, due to local rules … competence … ceasefires … getting jailed for doing normal work in China … DoD tenure? …

• Why am I not worried that we’re about to run out of AI progress to keep fueling the current capital inferno blazing around the world? The same week that Google dropped Gemini 2.5 and OpenAI released a buzzy new image generation technique, Databricks announced a new “model tuning method” that it calls “Test-time Adaptive Optimization (TAO).”

  • Databricks writes that “large language models are challenging to adapt to new enterprise tasks,” in part because “-tuning requires large amounts of human-labeled data that is not available for most enterprise tasks.” Enter TAO.

  • What is it? Using only “unlabeled usage data” from companies, TAO “leverages test-time compute (as popularized by o1 and R1) and reinforcement learning (RL) to teach a model to do a task better based on past input examples alone, meaning that it scales with an adjustable tuning compute budget, not human labeling effort.”

    • Bullish!

Everyone wants a piece of the stablecoin boom

I had planned an update to our continuing stablecoin coverage today concerning work by World Liberty Financial — a Trump-affiliated crypto company — to launch a stablecoin. Called USD1, the token will be “100% backed by short-term US government treasuries, US dollar deposits, and other cash equivalent,” WLF writes, adding that “USD1 tokens will be minted on the Ethereum (ETH) and Binance Smart Chain (BSC) blockchains, with plans to expand to other protocols in the future” while “USD1 reserves will be custodied by BitGo.”

That’s quite a nest of potential conflicts of interest for a company that POTUS launched before he retook office. Apart from the plague-level ick factor in watching a President work to cash in on their own policy choices, the move is interesting:

• Stablecoins are useful, offering consumers around the world access to the USD without the need to go through traditional channels to purchase, hold, or spend the global reserve currency.

• But while stablecoins are popular in both American and Web3 circles, they have thus far been third-party efforts. Think names like Tether and USDC, the latter of which is helmed by a private company that intends to list domestically.

• The fact that domestic stables are not centrally-run is a feature not a bug. Indeed, the concept of creating central bank digital currencies (CBDC) is considered anathema here in the States. So much so that Trump’s January crypto EO tried to block CBDCs, “which threaten the stability of the financial system, individual privacy, and the sovereignty of the United States, including by prohibiting the establishment, issuance, circulation, and use of a CBDC within the jurisdiction of the United States.”

  • The same Trump EO defines a CBDC as a “form of digital money or monetary value, denominated in the national unit of account, that is a direct liability of the central bank.”

• My question is simply this: If POTUS says that the government cannot make a stable, and then a crypto company affiliated with his persona and politics creates a stablecoin, how is that not a CBDC in a different wrapper?

Rules for thee, not for me.

The CBDC debate isn’t academic, mind. The EU is making noise that, per Bloomberg, “Europe needs a digital currency to safeguard against threats from new forms of money like stablecoins.” That’s according to an EU economist!

Meanwhile, the new Trump stable will face not only existing competition, but new names as well. While Tether and Circle are private — and in the case of the latter, venture-backed — Fidelity, which is not, is also jumping into the stablecoin arena.

I’d hazard that once there’s a super-solid mixer built to swap stables in a decentralized fashion we’re going to see the distinction between the various dollar-equivalents fade. But, today, with interest rates above zero, holding onto a bunch of cash for users is lucrative as hell. So, in go market participants until the current margins get zapped by competition.

Sovereign Servers

Yesterday CO noted that EU cloud companies are having a moment. Wired wrote that Elastx and Exoscale were seeing “an uptick in potential customers looking to abandon US cloud providers over the last two weeks.” That felt like a trend, so I went and looked at both companies this morning.

Elastx, a Swedish cloud computing company, wrote in its 2024 review that after “a longer period with a number of heavy strategic investments and red numbers,” it racked up “good growth during the year and turned to a positive operating profit which will land around 0 for the year.” Elastx went on to state its 2024 results put it at a point “where the curve points to record-breaking growth and a good operating profit” in 2025.

• Elastx is backed by Sobro, which reports a stake of just over 50%, and EUR47 million in 2023 revenue for its investment.

The company is beating the drum of its European foundation, writing in a recent post that the "cloud computing “landscape is changing rapidly, as AI and the rapid political swings in the US and Russia impact the entire industry” (machine translation). That’s an understatement.

Exoscale — owned since 2017 by A1 Digital, is in turn owned by A1 Telekom Austria Group — is a bit harder to chase down in results terms, but also flexes heavily on its EU roots. Hell, the company self-describes as a “solid European cloud hosting alternative” on its landing page, for heaven’s sake.

That these two companies which were seemingly doing fine-to-good before they caught a tailwind are now going to grow faster is great news for Europe. Faster growth means more investment, more investment means more bloc compute capacity, allowing for spend to circulate locally instead of being partially shipped off to Seattle, Redmond, or Mountain View.

And I think the move by EU companies to shift to EU options for traditionally USA-provided cloud services could provide a boost to other technology fields. AI models, for example. Wouldn’t Mistral benefit from the same vein of EU-level unity? I suppose you could write unity there as risk intolerance for American political risk, but you can pick your phrasing.

• Risk? What risk? Well, imagine that there’s a scrap between the EU and Trump. What is a lever that an American president could pull to try and lever EU folks to our will? How about access to American tech companies? This could be done in a number of ways, including something as simple as increased tax levies, or a demand that certain cutting-edge offerings are not allowed. I’d call myself paranoid, but currently the United States is trying to steal Greenland from Denmark. We’re in weird, stupid times domestically. Which could give Europe a true shot at building larger tech champions for itself.