The shoe that's been dangling for two years has finally dropped.

On June 25, OpenAI officially opened Sora video generation to all ChatGPT Plus users. Not a waitlist, not invite-only beta, not cherry-picked demos -- a real product that any ordinary user paying $20/month can open ChatGPT and use.

The significance of this day rivals the original ChatGPT launch: it marks AI video generation officially crossing the "usability threshold." For the past two years, AI video tools were stuck in the "generate a few seconds of eye-candy clips for social media novelty" phase. Today's Sora is a tool that can support actual production workflows.

What Does Sora Bring?

Every number on the production Sora spec sheet is rewriting industry expectations:

It supports 1080p resolution and up to 60 seconds of video generation -- already covering the vast majority of short-form ad duration needs. More critically is cinematic camera control: users can specify push-in, pull-out, pan, tracking, and other camera movements; AI generates smooth camera motion following cinematic language, not the stiff jitter of "the picture is moving but there's no cinematography" that plagued earlier versions.

Character consistency is another qualitative leap. One of the biggest pain points of AI video was "the same person's face keeps changing within a shot," making it unusable for storytelling. Production Sora supports reference-image character locking; generated characters maintain high consistency in facial features, clothing, and posture throughout the full 60 seconds -- meaning AI can finally shoot "stories with people in them," not just landscapes and B-roll.

Additionally, Sora supports mixed text+image input: you can upload a product photo and have AI generate video around it, or upload storyboard sketches and have AI generate dynamic footage in your visual style. This capability is nuclear-grade for advertising and e-commerce -- because it means "using AI to edit/iterate on footage" becomes possible, not just "generating from scratch."

China's Big Three: Fiercer in Video Than Text

The same week Sora opened up, Chinese AI video teams were also dropping major moves, the timing almost looking coordinated.

Kuaishou's Kling AI crossed 10 million daily active users, becoming the highest-DAU AI video product globally. Kling's edge lies in its understanding of Chinese-language context and Chinese user needs -- the quality of generated Chinese scenes, Asian faces, and localized storylines has surpassed Sora in many benchmarks. More importantly, Kling is deeply embedded in the Kuaishou ecosystem; users can publish generated videos directly to Kuaishou, creating a "generate-distribute-monetize" closed loop.

ByteDance's Jimeng AI announced 4K resolution video generation the same week, making it the world's first product to mass-produce 4K AI video. 4K means AI-generated video reaches professional broadcast standards for the first time -- not just good enough for Douyin, but good enough for TV ads, outdoor billboards, and theatrical pre-roll.

Vidu (Shengshu Technology) took a differentiated route: real-time video generation. Users type text in the chat box and video starts playing almost simultaneously, with latency under 3 seconds. This "real-time interactive" experience transforms video generation from "rendering wait" to "conversational creation" -- imagine telling AI "dim the light," "have the protagonist turn around," "change the background to the beach," and the video changes instantly, like talking to a director.

A noteworthy phenomenon: in the large-model text race, Chinese teams have always been playing catch-up; but in video generation, the gap between Chinese teams and OpenAI is minimal, and in some dimensions (DAU, 4K, real-time generation) they've already pulled ahead. The reason is simple -- China has the world's largest short-video market, the most mature creator ecosystem, and the most intense content competition, and demand-side pressure is forcing video generation technology to iterate faster.

The Cost Curve Gets Slashed: Earthquakes in Three Industries

What does AI video crossing the usability threshold mean? The most immediate impact is an order-of-magnitude drop in content production costs.

Advertising: A 30-second brand product spot traditionally required planning, scripting, casting, set building, shooting, and post-production; budgets ranged from 50K to 500K RMB with a one-to-two-week cycle. Now, using Sora or Kling, a skilled creator can generate 10+ candidate videos in 2-3 hours, pick the best versions for polish, and keep total cost under 500 RMB -- a 100x cost reduction, cycle compressed from two weeks to half a day.

E-commerce: Product hero videos and detail-page videos are table stakes in e-commerce, but shooting a product video used to require set design, lighting, models, and editing -- at least three days and thousands of RMB. Now upload a single product photo, and AI generates multiple versions -- 360 showcase, contextual lifestyle, usage demo -- in under 3 minutes at near-zero cost. Taobao and Douyin merchants are already replacing traditional shoots with AI video at scale.

Short-form video creation: Creator workflows are shifting from "shoot-and-edit" to "prompt-and-edit." A short-form creator's typical workflow used to be: write a script, prep props, shoot, import into editing software, add subtitles, score, color grade, publish -- a high-quality short took 4-8 hours. More and more creators' workflow now is: write prompts, generate video footage, AI auto-edits with subtitles, tweak and publish -- the whole process compressed to under 30 minutes. A creator's core competitive edge is shifting from "shooting and editing skill" to "storytelling ability and aesthetic judgment."

Not Everyone Is Happy

Every productivity tool revolution creates winners and losers simultaneously. The mass adoption of AI video is no exception.

Film/TV post-production is taking the first hit. VFX shots, scene extensions, and footage repair that used to take an entire post team weeks can now be done by one person with AI tools in hours. Low-to-mid-tier post-production outsourcing business is visibly shrinking.

Animators face an even deeper impact. AI video generation is essentially reconstructing visuals after "learning" from massive animation and film footage; 2D animation, Motion Graphics, and simple 3D animation are being rapidly displaced. The moat that independent animators relied on -- "one person making a short film" -- is being leveled, because now anyone can "one-person a short film."

The modeling industry is already feeling the earthquake. E-commerce models, print models, and even some commercial ad models are seeing work replaced by AI-generated virtual models -- no fees, no scheduling conflicts, no diva behavior, adjustable facial features and body types on demand, precisely matched to any target demographic's aesthetic preferences.

Video Is a Bigger Market Than Text

One bigger judgment to close.

For the past three years, the AI spotlight has been almost entirely on large-model chat -- ChatGPT, Claude, Gemini, Doubao, Tongyi Qianwen, with valuations and funding all centered around "AI conversation." But an overlooked fact: over 70% of internet traffic today is video, over 60% of global digital ad spend goes to video formats, and user time on short-video platforms is over 10x that of purely text-based platforms.

Video is a bigger market than text. The commercial value of AI video generation could ultimately be larger than large-model chat -- because it cuts directly into proven, massive paying markets (advertising, e-commerce, film/TV, education, gaming) rather than creating a new demand from scratch.

OpenAI clearly sees this, which is why Sora wasn't released as a standalone product but deeply integrated into ChatGPT as part of the workflow. China's Kling, Jimeng, and Vidu see it too, which is why they're pushing productization and commercialization faster than text models did.

Summer 2026 may well be remembered as the real starting line for AI video. Just as ChatGPT's launch in late 2022 wasn't the end of large models but the beginning, Sora's full release is only the first starting gun of the video generation era. Over the next 12 months, more players, more products, and more use cases will surface; the barrier to video creation will keep dropping, until "making a video" is as simple as "writing a sentence."

See you tomorrow.