Justifying the trillions

Eye-watering spending by AI giants may not necessarily make our lives better. In the meantime, Anthropic's CEO says "powerful AI" is on the way.

Read from your browser The Prompt Welcome to The Prompt, your Monday dive into the world of AI that puts Asia front and center. From Big Tech's power players to the region's scrappy disruptors, we cover it all. Want full access to all our AI reporting? Subscribe to us here. Hello Betamax, Just when you thought the checkbooks couldn't get any bigger, Masayoshi Son is reportedly ready to double down on his OpenAI bet. SoftBank is said to be in advanced talks to lead a new funding round for the Sam Altman-led company. The Japanese investor's participation alone comprises US$30 billion, according to reports last week. These eye-watering numbers might not make sense to us, but it seems like investors are buying it. Meta recently said it plans to spend US$135 billion in capital expenditure this year. Investors are also loving the news, with the Mark Zuckerberg-led company's stock climbing around 10% right after the announcement was made. But even as AI titans continue to pull in big bucks, questions are mounting over whether their products can justify the spending. There's a chorus of critics getting louder, pointing out that Silicon Valley is spending hundreds of billions of dollars on infrastructure for LLMs and agents that don't generate the revenue to match these investments. So these companies need to suck more money out of us. This is evident in OpenAI's plans to integrate ads into ChatGPT, even if Altman himself said just over a year ago that ads were a "last resort" for them, income-wise. This also brings us to agentic commerce, the subject of today's top AI read. My colleague Scott notes that experts are predicting a rapid rollout where OpenAI and Google transform their chat windows into checkout counters. Major retailers like Walmart, Target, and Shopify are already on board, integrating their systems to allow AI agents to handle orders directly. Meanwhile, Anthropic CEO Dario Amodei provided a sharper warning last week by suggesting that "powerful AI" could come to us soon, and they could morph into a "country of geniuses in a data center." With hundreds of billions at stake, OpenAI's shift toward advertising, the huge bets on agentic commerce, and expansive visions of "powerful AI" underscores the growing pressure to make the economics of AI add up. The question is whether any of this will actually make our lives better. Miguel Cordon, journalist TOP AI READ FROM OUR DESK Add to bot: how AI agents could reshape online retail By offering transaction fees as low as 4%, AI-powered retail services like Google's UCP and OpenAI's Instant Checkout provide a lucrative alternative for merchants. In comparison, Shopee's typical fees range between 5% and 15%. For now, the US is leading adoption with partners like Walmart and Shopify, although experts say that it'll be easy to bring these services to Southeast Asia. But would it be worth it for SMEs in the region to change the status quo? They already take advantage of the robust delivery and payment infrastructure of players like Shopee and Lazada, which are hard to replicate as a small business. WHAT WE THINK 1️⃣ Why US$1.5 billion is pennies for Anthropic Why would Anthropic even want to scan all the books in the world? A shocking revelation from The Washington Post suggests that books can teach its models "how to write well," as the open internet is apparently too low-quality. This was revealed in filings from the class-action lawsuit filed by a group of authors that alleged Anthropic of infringing copyright. This led to the company paying US$1.5 billion as a settlement. This looks like a setback for Anthropic, but the settlement could still be considered a win for the company. While Anthropic paid up for how it acquired some of the books (piracy), a judge ruled that the act of training AI on them was "quintessentially transformative." So, provided the company buys the books to train its AI on, it's fair game. But that doesn't feel right. While yes, the act is legal, the fact that Anthropic is a giant corporation that has a product that hinges on the work done by these actual human writers has to matter here. Compared to the US$70 billion in revenue Anthropic is eyeing by 2028, what these authors make is minuscule. And with the valuations and funding rounds that AI companies have grown accustomed to, US$1.5 billion is chump change. 2️⃣ Hardware bets are off China has reportedly approved the first batch of imports for Nvidia's H200 AI chips. This stance shift was made to help meet the compute needs of Chinese companies. But there's a catch: China still requires local firms to reach a quota of domestic chip orders before getting approved to purchase Nvidia's hardware. Right now, the US gets the tax revenue, Nvidia gets the sales, and China gets the computing power. Win-win-win. But everyone knows that the moment China's domestic chips are "close enough," the country will kill the Nvidia deal. Or the moment China gets too close to the "frontier," the US will pull the plug and dangle its top-shelf hardware in front of its Eastern neighbors once again. It's just a matter of what happens first. STANDOUT AI DEALS 1️⃣ AI butler IrisGo nets $2.8m, eyes series A after CES buzz The US startup is preinstalling its software on computer giant Acer's hardware, and is looking to hit a million daily active users by the end of the year. Unlike Copilot, which comes bundled with Windows machines that have Microsoft's 365 apps installed, IrisGo argues that its software moves beyond generating text. Instead, the AI butler operates the hardware itself, being able to record screens and control a device's mouse and keyboard. 2️⃣ A47 secures $2m to build AI-native news platform In contrast to the media industry's somber mood, A47 is projecting a bold vision: to build a news infrastructure layer that publishers and creators can trust. WHAT'S BUZZING "This is the single most powerful, destructive technology in human history," posted by @AlexFinn on his creation, ClawdBot. Finn positions ClawdBot - now Moltbot - as an AI assistant that can act as a virtual employee of sorts. Anthropic sent a cease and desist order to the project's creator regarding the old name's similarity to its Claude model. Yes, it does much more than a typical agent, but it also expects users to give an autonomous AI free rein over their personal systems. Arguments on if it's worth it or not ensued. ADVERTISE | SUBSCRIBE | HIRE | FIND JOBS P.S. Don't miss out on the biggest tech news and analysis. Add newsletter@techinasia.com to your address book, contacts, or safe sender list. Or simply move us into your inbox. Tech in Asia thanks our partners for supporting our newsletters. Sponsorship has no influence on editorial content. Too many emails? Switch to a different frequency or get new content through our preference center, or unsubscribe. You can also remove yourself from all Tech in Asia emails over here. Copyright © 2026 Tech in Asia, All rights reserved. Tech in Asia (Podium Level 3), 1000 Toa Payoh North, News Centre, Singapore 318994 A member of The Business Times.