AI bubble grows as bets rise

Voice AI takes off, China's AI tigers eye public markets, and a 22-year-old challenges how models are built. AI's next phase is unfolding.

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, The AI bubble hasn't burst - at least not yet. If this week's stories make one thing clear, it's that there is a real fight for AI inside labs, model architectures, and IPO pipelines. Take ElevenLabs. The voice AI startup's CEO recently told us that his company just crossed US$300 million in annual recurring revenue barely a few years after launch, a pace most SaaS companies would kill for. The company isn't stopping at audio. Its push to combine voice and video generation hints at fully AI-made content that no longer needs stitching across different tools. ElevenLabs' valuation rose to US$6.6 billion at breakneck speed, proving that there's huge demand for its services. Nvidia's backing could further accelerate its expansion, giving it better GPU access, more favorable pricing, and the infrastructure support needed to scale. China is gearing up for its own proving ground. Two of its most valuable model developers, Zhipu AI and MiniMax, are preparing to go public. They are unprofitable and burning capital, but they're racing to go public anyway. Why? Because the real contest isn't who builds the biggest model, it's who convinces the public markets that it can monetize one. We also profiled a 22-year-old NUS student who is asking whether the transformer architecture itself needs a reboot. That line of thinking led him to becoming an AI research scientist in a Silicon Valley VC firm. Meanwhile, OpenAI can't afford stepping out of the spotlight. It projects a leap from 35 million to 220 million paying ChatGPT subscribers by 2030. Different fronts, one takeaway: the AI bubble isn't here just yet. Samreen Ahmad, journalist TOP AI READS FROM OUR DESK 1️⃣ AI voice generating startup ElevenLabs crosses $300m in ARR ElevenLabs shows where AI is heading next: natural, human-sounding voice. Beyond English voices, the company has optimized its solution for Japanese, Korean, Malay, Thai, and Vietnamese, signaling that voice AI is no longer a Western-only product. Its licensing model, where people can upload AI versions of their voices and earn when others use them, also tackles one of the industry's biggest issues - consent and ownership - instead of avoiding it. 2️⃣ Decoding China's genAI duel as it shifts toward IPO arena Zhipu AI and MiniMax lining up for listings marks a turning point for China's AI sector. Until now, most of the action has been in state-backed funding rounds and not in public markets - where performance, transparency, and revenue matter. This development also points to how China wants its "AI tigers" to step onto a global stage and not remain domestic experiments. 3️⃣ NUS student joins OpenAI co-founder to rethink how AI models work A Singaporean undergraduate landing in a Silicon Valley VC lab is unusual on its own, but what stands out is the mandate of rethinking the foundations of AI, not just improving margins. If Mathew Vanherreweghe and the rest of Pebblebed Ventures succeed, it could lower the cost of future AI models and open the field to players beyond the billion-dollar giants. WHAT WE THINK 1️⃣ Will we still browse next Christmas? The holiday season is upon us, and by next Christmas, we may not be scrolling through product pages at all. We might be shopping through an AI assistant. OpenAI has rolled out a new ChatGPT feature called shopping research, which generates detailed buying guides. For now, it is focused on product discovery and comparison, not on completing purchases end-to-end. But OpenAI's endgame may be to turn ChatGPT into a universal personal commerce agent, one that helps discover, evaluate, and eventually buy products seamlessly. 2️⃣ Nvidia goes defensive for the first time Nvidia's response to Google's fast-rising AI chips shows how the battle for the future of AI hardware is heating up. Controlling between 70% and 95% of market share today, Nvidia doesn't just want to lead - it wants to make clear that GPUs, not TPUs, remain the industry standard. But as Meta considers spending billions of dollars on Google's chips, and with models like Gemini 3 being trained without Nvidia's chips, the landscape looks less one-sided than a year ago. FOUNDER FOCUS This week, we had a fun chat with Yashraj Bhardwaj, co-founder of Petonic AI, an AI-powered platform that helps businesses turn data into action. What was the "this is going to work" moment for you with Petonic AI? For me, it was the first time a Fortune 500 innovation team used our flagship platform SolvAI to compress a three-week consulting cycle into 17 minutes. The team then asked if it could deploy SolvAI across its global offices. What is the most surprising or amusing behavior you have seen from a user? It is surprising to see how quickly large enterprise teams start treating SolvAI as a real colleague. In one global FMCG client, its R&D and strategy units began referring to SolvAI as "our fastest team member" because it could cut a multidepartment analysis cycle from weeks to hours. If you could have any global CEO or public figure adopt Petonic AI tomorrow, who's the dream adopter? Nvidia CEO Jensen Huang, because he embodies the intersection of imagination, deeptech, and scale - exactly where Petonic AI thrives. If Petonic AI could "train" one real-world system instantly, what would you choose? Public policy execution. If governments could implement evidence-based decisions with the same precision as our models, we could solve environmental waste, digital governance, healthcare, and education gaps within years instead of decades. If you could jump into any other startup for 24 hours to borrow one idea or skill, which company would it be, and why? Definitely OpenAI to borrow its culture of extreme velocity. The way it collapses research cycles, ships fast, and rethinks org-design around intelligence is something every frontier tech company should learn from. If your team had a mascot in the office, what would it be holding? Our team mascot would be a sleek metallic panther holding a holographic blueprint. The panther reflects speed, precision, and bold decision-making - how we want enterprises to operate with Petonic AI. And the holographic blueprint represents the future: always evolving, always being designed in real time. Catch up on past issues of the newsletter here. In the meantime, if you have any feedback or ideas, feel free to get in touch with Huong, our managing editor, at huong@techinasia.com. 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 © 2025 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.