SetupAI
繁
SourceStalecollected in 2h

General Atlantic Leads Funding for China’s Kling AI

Read original on Bloomberg Technology
#generative-ai#funding#video-gen

Major funding for a top-tier video AI model aiming for global market competition.

30-Second TL;DR

What Changed

General Atlantic is in discussions to lead Kling AI's funding.

Why It Matters

Securing a major US backer like General Atlantic could provide Kling AI with the resources to compete globally against Sora and other video models.

What To Do Next

Benchmark Kling AI's video generation capabilities against current SOTA models like Sora or Runway Gen-3.

Who should care:Developers & AI Engineers

Key Points

  • •General Atlantic is in discussions to lead Kling AI's funding.
  • •Kling AI is the video generation arm of Kuaishou Technology.
  • •The funding aims to strengthen the company's position ahead of an IPO.
Key numbers$2 billion$18 billion$240 million$500 million

Deep Insight

Background and context from public sources — not the original article. 17 sources cited.

Enhanced Key Takeaways

  • •Kling AI, a video generation unit of Kuaishou Technology, was initially launched in June 2024 and has rapidly evolved, with its latest version, Kling 3.0 Omni, released in February 2026, offering native 4K resolution at 60 frames per second and audio generation in six languages.
  • •The company is seeking to raise over $2 billion in this funding round, targeting a post-investment valuation of $18 billion, which represents approximately two-thirds of Kuaishou's current market capitalization, signaling strong investor confidence in its growth potential.
  • •Kling AI has demonstrated significant commercial success, achieving an annualized revenue run rate (ARR) of $240 million by December 2025, which further surged to approximately $500 million by March 2026, driven by both enterprise API usage and consumer subscriptions.
  • •Kuaishou Technology views Kling AI as a core component of its dual AI strategy, focusing on strengthening its platform and creating new revenue streams, and has committed substantial capital expenditure, including RMB 26 billion (USD 3.6 billion) for 2026, primarily for AI computing infrastructure.
  • •Kling AI has seen broad adoption, with over 60 million creators globally using its products by the end of 2025, generating more than 600 million video clips, and is being integrated into professional workflows across advertising, e-commerce, film, and television industries.

Competitor Analysis

Kling AI 3.0 Omni (Kuaishou)
Key Features
Native 4K @ 60fps, audio in 6 languages, multi-shot narrative, physics-accurate motion, image-to-video, video extension.
Pricing (as of 2026)
Starts at $6.99/month. API: $0.0294/sec (Standard), $0.1029/sec (Pro).
Benchmarks/Notes
Elo score 1,248 (Artificial Analysis). Known for cinematic clips, but single-model lock-in and quality degradation on long extensions.
Seedance 2.0 (ByteDance)
Key Features
High motion quality, multi-shot storytelling.
Pricing (as of 2026)
Not specified.
Benchmarks/Notes
Highest-rated video model (Elo 1,269) on Artificial Analysis.
Runway (Gen-4.5)
Key Features
Professional editing interface, per-frame motion consistency, Act-Two motion capture, extend, expand, modify tools.
Pricing (as of 2026)
Free tier (limited), paid tiers from $12 to $76/month.
Benchmarks/Notes
Sets visual quality standard, fewer artifacts than Kling, strong for professional creative control.
Sora 2 (OpenAI)
Key Features
Longest single-pass clips with physics accuracy, cinematic photorealism, narrative continuity.
Pricing (as of 2026)
High price and access barriers.
Benchmarks/Notes
Benchmarks near Veo 3.1 on independent evaluations for 1080p quality.
Google Veo 3.1
Key Features
Spatial audio, 4K upscaling, audio-inclusive clips, sophisticated camera control, enterprise-grade output.
Pricing (as of 2026)
Custom enterprise pricing.
Benchmarks/Notes
Outputs visibly sharper, more detailed video; strong frame consistency.
Hailuo AI (MiniMax)
Key Features
Fastest generation, strong emotion, good for social media content.
Pricing (as of 2026)
$9.99/month.
Benchmarks/Notes
Generates clips in under 2 minutes. Practical for high-volume social media.
Pika Labs
Key Features
Fast social content, viral creative effects, accessible.
Pricing (as of 2026)
80 free monthly credits (no watermarks).
Benchmarks/Notes
Strongest free tier among video platforms.
HeyGen
Key Features
Script-to-finished video with AI presenters, voiceover, narrative structure.
Pricing (as of 2026)
Not specified.
Benchmarks/Notes
Best for finished, presenter-led videos, product demos, training content.
LTX Studio
Key Features
Complete production workspace, multi-scene projects with consistent characters, keyframe-level precision.
Pricing (as of 2026)
Not specified.
Benchmarks/Notes
Built for professionals needing script-to-finished video in one place.

Technical Deep Dive

  • Architecture: Kling AI utilizes a diffusion-based transformer architecture (DiT) enhanced by Kuaishou's proprietary 3D variational autoencoder (VAE) network.
  • Spatiotemporal Compression: The 3D VAE network enables synchronous spatiotemporal compression, which processes spatial relationships (object appearance) and temporal relationships (object movement) simultaneously. This improves video quality and training efficiency.
  • Attention Mechanism: A computationally efficient, full-attention mechanism serves as the spatiotemporal modeling module, allowing the model to accurately capture complex motion and details, including fast-moving objects and drastic scene changes.
  • Kling O3 (Omni One): The latest iteration, Kling O3, employs a unified multimodal architecture called "Omni One." This framework integrates understanding, generation, and editing of video content. It features 3D Spacetime Joint Attention combined with Chain-of-Thought reasoning to understand full 3D space and time relationships simultaneously, leading to physics-accurate motion and consistent character identity across shots.
  • Output Capabilities: Kling 3.0 Omni generates native 4K resolution at 60 frames per second, a significant leap from previous versions that topped out at 1080p and 30fps. It also includes native audio generation (dialogue, sound effects, ambient sound) in 6 languages.
  • Input Methods: Supports text prompts, reference images, and existing videos as input, maintaining consistency across these modalities.
  • 3D Face and Body Reconstruction: Kling 1.6 incorporates deep learning-based 3D modeling to reconstruct facial structures, expressions, and body postures from 2D inputs, enhancing realism with lifelike depth and motion continuity.

Future ImplicationsAI analysis grounded in cited sources

Kling AI's planned IPO in Hong Kong by early 2027 will likely solidify its position as a leading independent AI video generation entity.
The current funding round is explicitly viewed as a pre-IPO financing, with a substantial valuation target, indicating a clear path towards an independent public listing to capitalize on its rapid growth and market traction.
General Atlantic's investment could encourage more Western venture capital into Chinese AI companies, potentially easing geopolitical investment tensions.
As a prominent US private equity firm, General Atlantic's decision to lead funding for Kling AI is a notable move in China's competitive generative AI space, especially given the high tensions between Beijing and Washington over technology.
Kuaishou's continued heavy investment in AI infrastructure, despite short-term profit impacts, will likely accelerate Kling AI's technological advancements and market dominance.
Kuaishou is significantly increasing its capital expenditure for 2026, dedicating a large portion to AI computing infrastructure, which is a strategic move to maintain competitiveness and drive innovation in the rapidly evolving AI video generation market.

Timeline

2024-06
Kuaishou launches the first version of Kling AI, making it available for public testing within its video editing app, KuaiYing.
2024-07
Kuaishou launches full beta testing for Kling AI to global users and introduces a subscription program for users in mainland China.
2024-12
Kuaishou releases Kling 1.6, featuring upgraded video generation capabilities.
2025-03
Kuaishou Kling AI integrates DeepSeek-R1 to assist users in generating or optimizing prompts for video and image creation.
2025-04
Kling 2.0 is released, followed by Kling 2.1 in May 2025, introducing different quality modes and a global launch for Kling AI 2.0.
2026-02
Kling 3.0 / Omni is released, offering native 4K resolution at 60 frames per second and synchronized audio generation.
Read original article on Bloomberg Technology

Weekly AI Recap

Read this week's curated digest of top AI events →

PostLinkedIn

AI-curated news aggregator. All content rights belong to original publishers.
Original source: Bloomberg Technology ↗

This is a summary, not the original. Read the source, or get the weekly briefing.

The weekly digest

One email a week. Unsubscribe anytime.