Welcome to the second edition of the Starring Nobody Intelligence Report - a monthly deep-dive tracking the data, deals, and fault lines as Hollywood navigates the most disruptive shift in its hundred-year history. Each edition covers the previous month's developments and speculates on where things are heading.

Story #1 - The Sora Collapse: On March 24, OpenAI shut down Sora. The AI video generator that was supposed to be Hollywood's future ran at a reported $15 million per day in compute costs while generating just $2.1 million in total lifetime revenue. Disney's reported $1 billion investment deal - built around licensing 200+ characters to train Sora - collapsed with it. The most-hyped AI product in Hollywood history lasted less than six months.

Story #2 - The "Tilly Tax": As formal SAG-AFTRA contract talks stalled on March 15 - with the master contract expiring June 30 - the union's most novel proposal became public: a "Tilly tax," named after Tilly Norwood, a prominent AI-generated actress. The proposal would levy a fee on synthetic performers equal to the cost of hiring a human actor. The logic: if AI talent costs the same as human talent, studios will choose human every time.

Story #3 - The Tool Race Reshuffles: With Sora gone, the AI video generation market reorganized around three competitors: Kling 3.0 (March 8 release - 4K output, 6-axis camera control), Veo 3.1 (synchronized audio generation), and Runway Gen-4.5. The AI video market reached $946 million in 2026. Google, which backs Veo, is now the only player with true scale in the space.

These are the key metrics forming our baseline - updated with March's developments. Changes are flagged below.

• Entertainment jobs at risk by 2026: 100,000+ (Animation Guild - unchanged)
• VFX employment decline vs. 2022 peak: ~40% (VES - unchanged)
• AI video generator market size (2026): $946M ↑ from $717M prior year
• Disney / OpenAI deal status: COLLAPSED ⚠️ Sora shutdown, March 24
• OpenAI Sora daily compute cost: $15M/day vs $2.1M total lifetime revenue
• SAG-AFTRA master contract expiry: June 30, 2026 - 90 days remaining
• WGA 2026 negotiations start: April 16, 2026 - begins this month
• Kling 3.0 quality improvement: 2.1x vs prior model (released March 8)
• Higgsfield AI film contest submissions: 8,752 from 139 countries
• India AI film submissions: 1,805 vs 1,041 from the U.S.

ChatGPT is insanely powerful.

But most people waste 90% of its potential by using it like Google.

These 1,000+ proven ChatGPT prompts fix that and help you work 10X faster.

Sign up for Superhuman AI and get:

Claim your free prompts

The biggest deal in AI's Hollywood history lasted less than six months. What its collapse reveals about the true economics of the AI video moment.

On March 24, 2026, OpenAI announced it would shut down Sora - its AI video generation product - citing unsustainable compute costs. The numbers, when they emerged, were stark: $15 million per day in infrastructure costs against a total lifetime revenue of $2.1 million. Sora downloads had fallen 67% between November 2025 and February 2026.

Disney immediately exited its deal. Under the reported terms, Disney had agreed to invest approximately $1 billion and license 200+ characters from its Disney, Marvel, Pixar, and Star Wars libraries to train Sora's video generation models. Fan-inspired videos were to be featured on Disney+. That arrangement is now null.

CEO Sam Altman said he felt "terrible" telling Disney CEO Josh D'Amaro about the shutdown personally. Altman framed the decision as compute prioritization: "it's always about compute." Something else was working better. Sora wasn't.

The Sora collapse does not mean AI video generation is failing - the underlying market grew from $717M to $946M in 2026. It means something more specific: OpenAI built a product Hollywood got excited about, inked a landmark deal, and couldn't make the unit economics work. The shutdown puts Google in an unexpectedly dominant position. Veo 3.1 is now effectively the only enterprise-scale AI video platform with Hollywood-tier infrastructure. Watch for reports of Disney in conversations with Google about Veo as a replacement partner.

Those numbers aren't a CX issue — they're a design issue. Gladly's 2026 Customer Expectations Report breaks down exactly where AI-powered service loses customers, and what the architecture of loyalty-driven CX actually looks like.

Get the report

Unable to ban AI performers, SAG-AFTRA has proposed making them cost the same as human ones. It may be the most consequential labor idea of the AI era.

Tilly Norwood is a fully AI-generated actress - a synthetic performer with no human counterpart - who gained significant public attention in early 2026. She has no agent, no residual demands, no union membership, no health insurance requirements, and no ability to say no to a role. She is available twenty-four hours a day. She costs a fraction of SAG-AFTRA minimum. She also gave the union's chief negotiator Duncan Crabtree-Ireland a name for their most creative AI proposal yet.

The "Tilly tax" would require studios to pay a fee - equal to the cost of a human performer at SAG-AFTRA minimum rates - whenever they choose to use a synthetic AI performer. It applies to both digital replicas of existing performers and to fully generated characters with no human counterpart. The logic, as Crabtree-Ireland explained: if synthetics cost the same as a human, studios will choose a human every time. Revenue from the fee funds union health and pension funds - the same funds being drained by declining residuals as AI displaces human work.

Formal SAG-AFTRA talks with AMPTP paused on March 15 without a deal. The master contract expires June 30, 2026 - approximately 90 days from now. The WGA negotiation begins April 16, creating a hard deadline that may either accelerate SAG-AFTRA talks or push them to dormancy until the writers settle.

Sora is gone. Three tools are competing to fill the void - and the winner may determine which company becomes the infrastructure layer of AI Hollywood.

Kling 3.0 (Kuaishou) - Released March 8 with 4K output, 6-axis camera control, and a 2.1x quality improvement over its prior model. The 6-axis camera control closes one of the major gaps between AI-generated footage and professionally shot content. Kling is the tool most likely to be used in the first commercially distributed AI-primary feature film.

Veo 3.1 (Google) - Launched February 2026. Generates synchronized audio - ambient sound, dialogue, and sound effects - in a single pass alongside video. For content where audio matters, this dramatically reduces the production pipeline. Google's infrastructure advantage makes it the only enterprise-scale player post-Sora.

Runway Gen-4.5 - Last major release November 2025. Gaining users post-Sora but hasn't shipped a major March update. Needs a refresh to stay competitive.

No major March announcements in AI music - which is itself the story. The automation of mid-tier scoring and library music has become background noise. Suno, Udio, and AIVA are now simply part of the production pipeline for streaming content, advertising, and YouTube. The session musician market for this work has not recovered, and there is no policy framework in active negotiation to address it.

Veo 3.1's synchronized audio generation adds a new dimension: AI video tools are now generating ambient soundscapes and incidental music as a byproduct of video generation. The content creator who previously would have licensed a Suno track to accompany their AI video may not need to anymore. The addressable market for AI music tools in short-form content is being cannibalized by the video tools themselves.

AI tools are being positioned as productivity multipliers within existing studios - with one skilled compositor using AI tools producing the work of five. That framing helps studios justify smaller headcounts. It does not change the outcome for the workers who used to be the other four. The Higgsfield AI film contest results (8,752 submissions from 139 countries) illuminate the global picture from the other direction: the tools that professional VFX artists fear as displacement engines are being used by a new generation of creators who never had access to professional VFX pipelines in the first place.

The Higgsfield AI global film contest drew 8,752 submissions from 139 countries with $500,000 in prizes. India alone submitted 1,805 films. The United States submitted 1,041. The highest-participation country is not a traditional filmmaking market. The democratization is real, global, and running ahead of the regulatory and labor frameworks being built to contain it.

While Hollywood's unions and studios are locked in a power struggle over AI's future role in filmed entertainment, Asia's screen industries - as demonstrated by the 28 AI-dedicated sessions at Hong Kong's Filmart in March - are rushing toward full embrace. The AI filmmaking ecosystem is not waiting for Hollywood to decide how it feels about AI. It is building a parallel industry.

→ WGA negotiations open April 16. AI provisions - training data consent, credit attribution, disclosure requirements - will be the central battleground. Watch for studio counterproposals on "AI-assisted" writing definitions.

→ SAG-AFTRA talks resume or stall. With WGA negotiations consuming oxygen, SAG-AFTRA may enter a holding pattern through April, then accelerate toward June 30. The "Tilly tax" will either gain traction as a genuine bargaining chip or get walked back as too radical for this round.

→ Watch for Disney-Google Veo reporting. Disney needs an AI video partner. Google needs premium IP. The logic of the Sora deal hasn't changed - the counterparty has.

→ Kling 3.0 adoption data will emerge. If Kling's March 8 release generates filmmaker community enthusiasm, it becomes the de facto standard for indie AI filmmaking.

→ Both SAG-AFTRA and WGA master contracts will have been renegotiated. The "Tilly tax" either appears as the most consequential labor innovation of the AI era, or was traded away for more conventional concessions on residuals and disclosure.

→ At least one commercially distributed feature film will carry a primary AI-generated credit. It will underperform critically, perform adequately commercially, and ignite the first serious public debate about whether AI-generated narrative content requires labeling.

→ OpenAI will have re-entered the AI video space in some form. The compute economics that killed Sora are a solvable problem. OpenAI will not cede this market permanently.

Scenario A - The Bifurcation (most likely): Hollywood splits into two tiers. Prestige content remains predominantly human-made. Below that, an ocean of AI-generated content fills streaming libraries, indistinguishable by most consumers from human-made work. The middle tier of entertainment employment is gone.

Scenario B - The Reclamation (raised probability post-Tilly Tax): Strong union contracts establish a functioning market for licensed human performance. AI becomes genuinely complementary. More content gets made. Some categories of employment shrink; others grow.

Scenario C - The Collapse (low probability, lowered by Sora shutdown): Studios largely exit the business of paying for human creative execution at most budget levels. The Sora shutdown makes this less likely near-term - compute costs of AI video at scale are a real constraint. But by 2031, compute is cheaper. The scenario remains on the table.

This is a premium subscriber report published by aiisgoingtokillyou.com. Sources: Hollywood Reporter, Variety, Deadline, Fortune, Bloomberg, Advanced Television, Art Threat (March–April 2026). Delivered on the 1st of every month.