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 reader,
Meta is reportedly building a cloud business to resell its excess AI compute capacity. It’s a strange fit for a company still burning billions on frontier models, until you realize that Meta is the only hyperscaler without a commercial cloud arm.
This isn’t Meta stepping back from the model race. Think of SpaceX’s compute playbook: sell the plumbing, not just the payload. For SpaceX, that approach already generates about US$2.3 billion in monthly revenue from just three customers: Anthropic, Google, and Reflection AI.
That’s the pattern we’re seeing in this week’s newsletter: amateurs sell tokens, pros sell control. Tokens get commoditized and cheaper every year, but control – whether it’s over compute, infrastructure, or data – doesn’t.
The winners of AI’s next phase won’t be the ones using flashy moves; it’ll be the ones who own the squares everyone else has to cross.
Image credit: Ulla
Palantir CEO Alex Karp made a version of this argument last week. He said that if intelligence is really as valuable as the AI labs claim, why charge by the token? A consultant who promises to make you a billion dollars wouldn’t bill by the hour – they would take a cut of the deal.
Karp was promoting Palantir’s partnership with Nvidia, allowing him to sell a direct alternative to what OpenAI and Anthropic offer. For sure, he has an incentive to delegitimize token pricing.
But the underlying point is worth noting. Our insights piece this week, written by Kyle Qi from Llama Ventures, focuses on a similar argument.
According to Qi, model prices have fallen roughly 1,000x in three years. Meanwhile, the layer behind the scenes – inference, orchestration, fine-tuning – is beginning to resemble infrastructure more than software. This explains why companies such as Fireworks, Together, and Baseten are commanding infrastructure-like valuations.
Elsewhere in this issue, Lightspeed venture partner Kevin Aluwi told us that adoption is Southeast Asia’s biggest AI bottleneck, not compute.
For our Founder Focus column, we sat down with Hiep Nguyen, COO of Vietnamese startup AI Hay. To monetize, it’s borrowing a playbook from other game developers. As to exactly what this means, you have to check out the conversation.
Different problems, same underlying question: Where do you build the layer others can’t easily replace?
TOP AI READS FROM OUR DESK
WHAT WE THINK
FOUNDER FOCUS
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.
