Realtime AI News
Kimi-maker Moonshot AI targets $2 billion in annual revenue
TechCrunch reports that Moonshot AI, the developer of the Kimi assistant, is targeting $2 billion in annual revenue. The same report cites OpenRouter data showing K3 models generating as many as 300 billion tokens a day, even as usage figures have slipped slightly in recent months.

TechCrunch reports that Moonshot AI, the company behind the Kimi assistant, has set a $2 billion annual revenue target. That number puts the commercial question front and center for a Chinese model lab competing in a crowded market.
The report pairs the target with a usage datapoint: OpenRouter, a third-party aggregation platform for model API calls, currently shows as many as 300 billion tokens being generated each day by K3 models. Token throughput on such platforms is widely watched as a rough proxy for how much a model is actually used.
The same report notes that K3's usage figures have declined slightly in recent months. That nuance matters: high token volume and durable revenue are not the same thing, and inference demand only becomes a business when it converts into paying customers.
A $2 billion annual target implies Moonshot needs to sustain a mix of consumer subscriptions, API consumption and enterprise deals. For most model companies, growing token usage does not automatically grow revenue — inference costs, free tiers and price competition all squeeze margins.
Kimi is Moonshot's main consumer-facing product, and K3 is the model family now absorbing much of that traffic. Making daily K3 throughput part of the public narrative suggests the company wants real usage scale to carry its commercial story.
Why it matters: the sector's scoreboard is shifting from benchmark results toward operating metrics. With API prices under pressure, what decides a lab's health is how many users actually pay for inference and whether that revenue covers compute costs.
What to watch next: whether Moonshot discloses a clearer revenue breakdown, whether K3 usage stops sliding, and whether figures like 300 billion tokens a day can be corroborated by independent measurement. The three curves — tokens, paid conversion and revenue — moving in the same direction is the real test.
Why it matters
If the target holds, Moonshot becomes one of the Chinese model labs under the most direct commercialization pressure. For the wider sector, the competitive scoreboard is shifting from capability benchmarks to the match between token throughput and paid revenue.
Nearby Updates
All09/12, 02:02
MYbank opens Bailing 2.0, a small-business finance Agent, to 42 million merchants
At the 2026 Bund Conference, MYbank disclosed its AI banking rollout for the first time: Bailing 2.0, described as the world’s first inclusive-finance Agent for small and micro businesses, is now open to 42 million merchants. Behind it, eight AI workbenches have moved into risk control, manual review, marketing and R&D, with AI coding doubling and 15% of business change requests completed by AI.
09/12, 00:46
Nscale adds former OpenAI exec Fidji Simo to its board ahead of potential IPO
TechCrunch reported on September 11 that Nscale has added former OpenAI executive Fidji Simo to its board as it prepares for a potential IPO. Simo was the No. 2 executive at OpenAI and previously led Instacart through its 2023 IPO.
09/11, 22:05
Anthropic Posted a $450K Sales Role Devoted Solely to Meta — Then Closed It in a Week
Anthropic briefly listed a job titled Mega Account Executive, Meta, a role serving only Meta, with on-target earnings of $380,000 to $450,000, or roughly 3.2 million yuan at current rates. The posting went live on September 3 and stopped accepting applications on September 9, as Meta engineers already lean heavily on Claude Code.
09/11, 18:08
Zhipu's GLM-OCR Model Update Tops OmniDocBench V1.5 at 0.9B Parameters
Zhipu's zai-org updated the GLM-OCR model card on Hugging Face on September 11, presenting a 0.9B-parameter multimodal OCR model that scores 94.62 on OmniDocBench V1.5 and ranks first overall. Built on a GLM-V encoder-decoder with a PP-DocLayout-V3 layout pipeline, it ships under the MIT license with documented vLLM, SGLang, Ollama and Transformers deployment paths.