Industry View · Audio-Visual
Generative video and audio crossed from demo to production line in 2025-2026. Netflix used AI footage in a finished show, Google's users generated over 275 million clips in Flow within five months, and ElevenLabs hit an $11B valuation on voice alone. The value is real, but it is pooling at the model layer and draining the stock-footage and dubbing floors.
ElevenLabs $330M+ ARR; Firefly ~$400M revenue; real revenue and cost displacement
Set design, wardrobe and storyboard exploration now happen as prompted video rather than commissioned art.
Text/image-to-video with synced audio is now a single API call, concentrating spend on a few foundation models.
Localization and narration collapse to software, with consent/residual fights unresolved.
AI fill, extend and de-age move from plugins into the core NLE, raising VFX without raising headcount.
Recommendation plus generative ad/creative variants tighten the loop between content and monetization.
01 · The thesis
The 2025-2026 inflection was synchronized audio. Once Google's Veo 3 (May 2025) and OpenAI's Sora 2 (Sept 2025) could generate dialogue, sound effects and ambient noise locked to picture, AV generation stopped being a silent-clip novelty and became a usable production primitive. Netflix's Ted Sarandos confirmed the company used generative AI in finished footage for The Eternaut, and de-aged characters in Happy Gilmore 2 — the first time a major studio admitted AI in shipped frames, not just pre-viz. The economics are brutal at the edges. The stock-footage industry, worth roughly $14B in 2019, has collapsed toward ~$3.2B, pushing Getty and Shutterstock into a $3.7B defensive merger. Voice is splitting in two: ElevenLabs cleared $330M ARR by end-2025 selling the very automation that threatens the 2M+ voice actors SAG-AFTRA spent an 11-month strike trying to protect. The tooling incumbents (Adobe) are racing to wrap models in 'commercially safe' licensing before the pure-play models commoditize them.
Set design, wardrobe and storyboard exploration now happen as prompted video rather than commissioned art.
Text/image-to-video with synced audio is now a single API call, concentrating spend on a few foundation models.
Localization and narration collapse to software, with consent/residual fights unresolved.
AI fill, extend and de-age move from plugins into the core NLE, raising VFX without raising headcount.
Recommendation plus generative ad/creative variants tighten the loop between content and monetization.
02 · The two clocks
Voice and dubbing is the fastest clock. AI localization already lets Netflix scale multilingual content, and a voice-recording producer estimated roughly a quarter of overseas clients expect to lose projects to AI automation — with 2M+ voice actors worldwide exposed, per Rest of World and The Wrap. Generation quality is compounding monthly. Google moved from 40M Veo videos in July 2025 to 275M+ Flow videos by October 2025, while Runway pivoted its $315M raise toward 'world models' rather than clips — the frontier is moving from seconds of footage to controllable simulated worlds. Stock media is the slow-motion collapse already finished. With the category down from $14B (2019) to ~$3.2B, the Getty-Shutterstock merger is consolidation under duress, not growth — the disruption here is largely complete, and the survivors are pivoting to AI data licensing.
03 · Public players & exposure
We plot the listed players on two editorial axes — how exposed each is to AI disruption, against how ready its data, brand and position are to be the answer. The figures in the table are sourced; the placement is our read.
04 · Private flagships
The companies attacking this industry AI-first, with disclosed funding where available:
From creative tool to 'world model' ambition, building simulators of physical reality rather than just clips.
Voice cloning, dubbing and narration at enterprise scale; explicitly building toward an IPO.
Turns training manuals into avatar-led video in 100+ languages for corporate L&D.
Sora 2 set the synced-audio bar, but the consumer app's April 2026 shutdown shows the format is still unsettled.
Wraps generation in IP-clean licensing inside Photoshop and Premiere, the incumbent's defensive moat.
Two disrupted incumbents merging to survive and pivot toward licensing data to AI trainers.
05 · Signals
06 · The exposure read
AI rewards clean, structured advantage and punishes friction. The line runs through who owns the data, the brand and the customer — and who is merely a step the technology can route around.
Sources
Disclosed valuations of AV-AI pure-plays, 2025-2026
Source: CNBC, TechCrunch, Forbes (2025-2026 funding rounds) ($B valuation)
Veo 3 adds synchronized audio
Google DeepMind ships text-to-video with dialogue, SFX and ambient sound, ending the silent-clip era.
Sora 2 and a social video app
OpenAI launches Sora 2 with synced audio plus an iOS app for remixing AI clips — later shut April 2026.
Netflix goes 'all in'
Q3 earnings confirm generative AI in finished footage and a broad push across content, ads and recommendations.
Getty-Shutterstock merger
A $3.7B combination signals the stock-media market consolidating under AI pressure.
Voice and video raises stack up
ElevenLabs ($11B) and Runway ($5.3B) close large rounds days apart, pricing the model layer richly.
From creative tool to 'world model' ambition, building simulators of physical reality rather than just clips.
Voice cloning, dubbing and narration at enterprise scale; explicitly building toward an IPO.
Turns training manuals into avatar-led video in 100+ languages for corporate L&D.
Sora 2 set the synced-audio bar, but the consumer app's April 2026 shutdown shows the format is still unsettled.
Wraps generation in IP-clean licensing inside Photoshop and Premiere, the incumbent's defensive moat.
Two disrupted incumbents merging to survive and pivot toward licensing data to AI trainers.
| Company | Stance | The sourced fact |
|---|---|---|
| Adobe ADBE | Incumbent pivot | Firefly generated over 22 billion assets within two years as of April 2025, and ships the 'commercially safe' Firefly Video Model into Premiere Pro. |
| Alphabet (Google) GOOGL | Model + distribution | Users generated over 275 million videos in Google Flow within about five months of its May 2025 launch, per Google (Oct 2025). |
| Netflix NFLX | All-in adopter | Ted Sarandos said Netflix used generative AI in final footage for the first time on The Eternaut, and is 'all in' on AI across content and ads (Q3 2025). |
| OpenAI OAI | Frontier model | Released Sora 2 with synchronized audio in Sept 2025; later wound down the consumer Sora app (shut April 2026) with no announced successor. |
| Runway RUNWAY | Gen-video pure-play | Raised a $315M Series E in Feb 2026 at a $5.3B valuation led by General Atlantic, with Nvidia, AMD and Adobe Ventures participating. |
| ElevenLabs ELABS | Voice leader | Raised $500M at an $11B valuation (Feb 2026) on $330M+ ARR; used by 41% of the Fortune 500. |
| Synthesia SYNTH | Enterprise avatars | Raised $200M at a $4B valuation; used by 90% of the Fortune 100 and crossed $100M annualized revenue, reportedly rejecting a $3B Adobe bid. |
| Getty Images GETY | Disrupted incumbent | Cited revenue decline of roughly 68% amid AI; pursuing a $3.7B merger of equals with Shutterstock while licensing data to AI firms. |
| Shutterstock SSTK | Disrupted incumbent | Cited revenue decline of roughly 71% amid AI substitution; merging with Getty to consolidate a shrinking stock-media market. |
| Disney DIS | Cautious tester | Tested Runway's tools and discussed potential use, but a spokesperson confirmed no current plans to integrate the software (2025). |