June8th, according to reports from Daily Economic News, AI unicorn Kimi (Moonshot AI) has officially kicked off its latest funding round, hitting a pre-money valuation of $30 billion, which is a massive leap compared to just last month.
For those keeping track of the money trail, Kimi is the flagship AI assistant under domestic AGI powerhouse Moonshot AI. By late 2025, the company had already wrapped up a solid $500 million Series C. Then, in mid-February this year, whispers circulated that another $700 million-plus deal was ready to close. Fast forward to March, and industry trackers noted Moonshot’s valuation had already climbed to $18 billion—essentially quadrupling in just three months while they lined up a fresh $1 billion round.
If you pull up the Tianyancha App, the investor lineup for that initial Series C was stacked with heavy hitters like IDG Capital, Alibaba, Tencent, and several other top-tier industry players. When that $700 million-plus February financing finally settled in 2026, the list expanded to include Alibaba once again, alongside Origin Capital, GaoRong Ventures, JD Health, and Tencent.
Just a few weeks later in early May, fresh headlines claimed Kimi was on the verge of closing a massive $2 billion round, pushing its post-money valuation past the $20 billion mark. At the time, we learned that Meituan’s Dragon Fund would lead the charge, with China Mobile and CPE (CITIC Private Equity) joining in. Dragon Fund alone reportedly pledged over $200 million.
Over the past few months, Kimi has seriously cranked up its product update pace. Just on June 3rd, Moonshot AI rolled out a public beta for Kimi Work, a brand-new local Agent built specifically for knowledge workers. It’s dropping alongside the latest test builds for both Mac and Windows clients. Here’s the cool part: you just describe what you need in plain language, and Kimi Work breaks it down on your machine, runs tasks in parallel, taps into external tools, navigates your browser, organizes your folders, and spits out polished documents, spreadsheets, PPT decks, and other work assets—all without you lifting a finger.
Looking back just a month earlier, on the evening of April 20th, Moonshot AI actually pushed out and open-sourced the Kimi K2.6 model. Mind you, that was only a couple of months after Kimi K2.5first hit the shelves. This beast is touted as their most powerful coding engine to date, with serious upgrades in long-context programming capabilities and a major boost in autonomous Agent execution. Kimi K2.6 is live right now on kimi.com, the latest Kimi app, via the Kimi API, and straight through the Kimi Code programming assistant. Seriously, everyone gets access.
Jump back to late January, when Kimi K2.5initially launched. The official drop highlighted massive leaps in multimodal processing. More importantly, leaning hard into the current Agent trend, they introduced both the standalone Kimi Agent and the highly anticipated multi- Agent coordination setup called Agent Swarm. This lets multiple AI Agents tackle complicated workflows side-by-side, really cementing Kimi’s shift toward fully agentic behavior.
It all started way back in November 2025 with the rollout of Kimi-K2 Thinking. They positioned it explicitly as an Agentic reasoning engine, and the early benchmark data proved it—major breakthroughs across both complex logical reasoning and Agent capability tests.
During the opening ceremony of the 2026 Zhongguancun Forum on March 25th, Moonshot AI’s founder and CEO, Yang Zhilin, shared some fascinating insights. He pointed out that open-weight models like Kimi K2.5 are quickly becoming the industry benchmark. Hardware makers now practically have to run their chips against these open-source tests to prove any real performance gains. Yang also noted a massive shift in how AI research is conducted. We used to lean heavily on scraping the web and relying on small teams of humans to manually label data for values and preferences. While reinforcement learning took center stage last year, humans were still doing the heavy lifting to pick high-quality training tasks—that’s why we saw such jumps in coding and math. But looking ahead? AI is taking the wheel. Researchers will soon be handed massive token budgets to automatically generate new tasks and environments. The models themselves will learn to define their own reward functions and even explore entirely new network architectures. With that loop in place, the whole AI R&D pipeline is going to speed up dramatically.
Then on March 30th, exclusive reports came to light revealing something pretty impressive: just one month after Kimi K2.5 went live, Moonshot AI’s ARR (Annual Recurring Revenue) had already shattered the $100 million mark. Insiders also mentioned that K2.5’s launch caused API throughput quotas (measured in TPM, or Tokens Per Minute) to get incredibly tight almost overnight. Demand was so hot that enterprise clients were literally dropping seven-figure upfront deposits and spending guarantees just to lock in priority server access.
Analysts at Changjiang Securities are tracking this closely, noting that institutional capital is flooding into AI at record speeds. As real-world AI applications start generating actual revenue, history tells us that infrastructure spending usually peaks about a year before the software boom hits. Based on that American playbook, AI apps are shaping up to hit a massive commercialization inflection point right around 2026.
A while back, rumors were swirling that Moonshot might be eyeing a Hong Kong IPO. When reporters tried to verify those claims back in March, their team politely declined to comment.
In fact, back in December last year right after that C round wrapped up, Yang sent out an internal memo that pretty much shut down any immediate public market speculation. He wrote: “The company recently completed a $500 million funding round and massively oversubscribed it, bringing our current cash reserves past 10 billion yuan.” He went on to note: “Compared to chasing listings on the secondary market, we firmly believe we can still pull in far larger sums from private investors. In reality, our combined Series B and C raises already dwarfed what most companies pull in during traditional IPOs or private placements. So yeah, we’re not in a hurry to go public, and frankly, it was never our end goal.”
To put things in perspective, look at what their rivals did. Both Zhipu and Minimax officially listed on the Hong Kong Stock Exchange in early January. Zhipu opened the doors at a staggering $58 billion HKD on day one, while Minimax briefly flirted with the trillion-HKD mark. Fast forward five months to today, June 8th, and the charts are wild: Zhipu’s market cap has exploded past $500 billion HKD, officially overtaking Baidu’s Hong Kong valuation, with shares smashing through the $1000 HKD ceiling; Minimax has similarly climbed past $150 billion HKD. These insane public market valuations are currently giving private equity investors a whole new set of metrics to work with.
