Moonshot AI has dropped its Kimi K3 model, a 2.8-trillion-parameter system that costs roughly half as much to run as comparable US models. The launch rattled Wall Street, sending the Nasdaq lower as investors recalibrated expectations for the AI arms race. It also reignited talk about decentralized computing as a potential counterweight to the industry's growing concentration.
A price-performance shock
Kimi K3 packs more parameters than most publicly known models from American labs — but the real surprise is the price tag. Moonshot AI claims the model can be deployed at roughly half the cost of equivalent US systems. That kind of cost advantage, if real, threatens the margins of companies that have bet big on expensive, proprietary AI infrastructure. The company didn't disclose how it achieved the savings, but analysts point to possible efficiencies in training or inference hardware.
Wall Street's jitters
The Nasdaq took a hit in the days following the announcement. Traders dumped shares of major AI-linked names, worried that cheaper competition could compress revenue growth. The selloff wasn't limited to pure-play AI firms; chipmakers and cloud providers also felt the pressure. The message from the market was clear: if a Chinese startup can deliver comparable power at half the cost, the premium pricing that many US AI companies have enjoyed may not last.
Decentralized computing gets a second look
The launch has also revived discussions about decentralized computing. Some developers argue that the cost advantage of models like Kimi K3 could accelerate a shift away from centralized data centers toward distributed networks that pool consumer hardware. Proponents say this would lower barriers for smaller players and reduce reliance on a handful of hyperscale cloud providers. Critics counter that decentralized systems still struggle with latency, security, and coordination at scale. But the conversation is no longer theoretical — Kimi K3 gives it a concrete benchmark.
The big question now is how US AI firms will respond. Price cuts? Faster model releases? Or a push for even larger, more capable systems that justify their current costs. The next few months will tell.
