If you can't beat 'em, host 'em

Why AI labs are hosting Chinese models to fund their frontier ambitions. Plus, we take a look at MiniMax’s revenue surge.

Read from your browser Daily Newsletter Welcome to Tech in Asia's daily newsletter, your essential dose of Asia's tech and startup buzz. Subscribe to this newsletter or send story tips to editors@techinasia.com. Ensure our articles appear in your Google searches: IN FOCUS For mid-market AI labs, Chinese models pay the bills MiniMax shows how free AI can also pay Hello Betamax, Four years after ChatGPT's launch, the dust is settling across the hundreds of AI labs worldwide. The market is breaking into clear buckets. At one end are US frontier labs backed by Big Tech: OpenAI, Anthropic, and Google. At the other are Chinese players such as Z.ai, Moonshot, and MiniMax that have reportedly distilled frontier capabilities into open models that anyone can download and host. Then come the middle powers. Companies like India's Sarvam and France's Mistral set out to build frontier models, often with a sovereignty pitch. If the Chinese players weren't in the picture, they might have sold their less-capable models to enterprises that did not need the most advanced US systems but wanted a cheaper, locally hosted alternative. But Chinese models can be downloaded for free and run anywhere, leaving few takers for the middle powers' offerings. Today's featured story examines their workaround: hosting Chinese models for enterprises on GPUs originally bought to train and serve their own AI models. This can buy AI labs time by giving them early streams of revenue as they build toward a frontier model. But for many of them, it could signal a complete pivot from model development to an inference cloud business. Pranav Balakrishnan, AI reporter TOP STORY Frontier dreams stall as labs turn to hosting Chinese models High-end GPUs, which do the heavy lifting, remain scarce and are typically secured through long-term purchases or leases. Though accessed through the cloud, capacity cannot easily be bought and sold on a spot market. That scarcity has encouraged buyers to lock up more capacity than they immediately need. Hosting Chinese models gives AI labs a way to put that reserved compute to work and generate revenue. But the business rests on a contradiction, as AI compute is scarce, even as much of it sits underused. That imbalance may not last. CAN AI COMMUNICATIONS SCALE? A curated roundtable luncheon by Sinch and Tech in Asia As AI becomes part of live customer journeys, maintaining reliable delivery and customer trust becomes more complex. Leaders now need to consider whether their communications systems can support growing automation without compromising the experience that customers receive. Held on September 15 as part of Tech in Asia Conference 2026, "Is Your Communication Infrastructure Ready for AI at Scale?" brings together decision-makers from across Southeast Asia to examine the practical realities of scaling AI-powered customer communications. Combining research-led insights with a moderated roundtable discussion, the session gives you space to raise questions relevant to your organization, compare approaches with peers, and continue the conversation over a three-course lunch. The conversation will cover how to maintain deliverability and channel reliability as automation grows, the role of email in an AI-first world, and the strategic considerations involved in protecting customer trust across increasingly automated journeys. This session is open exclusively to Tech in Asia Conference 2026 General, Pro, and VIP pass holders. Capacity is limited and all registrations are subject to approval. Already have a conference pass? Register your interest for the luncheon.] Not yet attending? Get your Tech in Asia Conference 2026 pass first. SPOTLIGHT How MiniMax makes money from free AI The momentum behind open models has become apparent in public markets. MiniMax, which is listed in Hong Kong, said its revenue nearly quadrupled to US$116.6 million in the first half of the year. MiniMax's business model resembles the one explored in today's featured story. The model weights may be free to download, but enterprises pay for the convenient version: access through the lab's own servers. Rather than download MiniMax's model and run it on their own GPUs, developers can send requests through its API, asking it to summarize a document or generate code. MiniMax runs the model, returns the response, and charges by usage. That makes open source less a giveaway than a distribution strategy. The free model draws developers in; the hosted, ready-to-use version generates revenue. MiniMax's growth points to the changing economics of the AI industry. OpenAI has struggled to meet its growth expectations this year, while leading frontier products such as Anthropic's Fable are facing a market crowded with free Chinese alternatives. WHEN AI PILOTS STALL, WHAT COMES NEXT? Join an exclusive breakfast session for AI leaders Moving from AI pilots to scalable enterprise capability remains a major challenge for enterprises. This 15 September, during Tech in Asia Conference 2026, Wonderful and Tech in Asia will bring together senior enterprise leaders for a focused discussion called "AI Leaders Breakfast: Closing the Enterprise AI Gap." It addresses why AI initiatives stall after early testing and examines the requirements across strategy, integration, governance, and organizational alignment needed for enterprise deployment. Expect candid peer exchange, focused discussion, and networking with leaders navigating similar execution challenges. This session is open exclusively to Tech in Asia Conference 2026 General, Pro, and VIP pass holders. Slots are limited, and all registrations are subject to approval. Have a conference pass? Register your interest for the AI Leaders Breakfast. PRESS PEEKS The following press releases were published as they were received, without any editing from the Tech in Asia team. You, too, can publish your own press release on our site by simply filling out this form. Our team will get back to you by the next business day. Assemblr EDU - Indonesia's education ministry partners with Assemblr and Samsung for nationwide visual coding championship: ImajiNation 2026 aims to engage 10,000 students and 1,000 teachers from more than 2,000 schools across Indonesia's 38 provinces. EVENTS HAPPENING You can also check out our curated list of trending tech events and Tech in Asia's signature events. Secure a startup booth at our conference Startup Booth, the exhibition space at Tech in Asia Conference serves as your high-velocity base to move beyond the pitch deck and into the production stacks of 1,500 decision-makers from companies like HSBC, Canva, Ferrero, and ComfortDelGro. For just US$750, you secure an exhibition slot for two days and two conference passes to bridge the gap from startup ambition to real-world results. Don't just showcase your product; start your distribution journey where Asia's tech future is built. Find out more here. WHO'S HIRING You can post a job here or search for jobs here. Marketing Manager (EduTech) at Cakap | Tangerang, Indonesia IT Governance, Risk, and Compliance (GRC) at PT. ASLI Rancangan Indonesia | Jakarta, Indonesia Digital Marketing Specialist at Boleh Dicoba Digital | Bandung, Java, Indonesia 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