AI Dramas in 2027: What Happens Next

Updated August 19, 2026 · Written by the AI Drama Reviews editorial team · 31 min read

DramaBoxEditor's pick — biggest short-drama library, lowest price to go unlimited 9.8Excellent

• Thousands of series, roughly 200 new titles every month
• Free episodes daily, plus ad-unlocks — no card needed to start
• Unlimited plans from about $5.99 a week, the lowest entry price we found
• Runs on Android, iOS and in a desktop browser

Go to DramaBoxAffiliate link · free to download · in-app purchases available

Affiliate link. We may earn a commission if you install through it, at no extra cost to you — it never changes our scores or placements. How we test · Disclaimer

The short answer, and the rule this page follows

There are exactly three kinds of statement anyone can make about AI dramas in 2027, and confusing them is the reason this category is so badly reported. A published forecast is a number a named research firm put its name to, on a stated date, for a stated scope. A trend is a claim about direction supported by evidence you can go and check. A guess is a judgement call that no evidence settles. All three are legitimate. Presenting the third as the first is not.

Every section below carries one of three labels. We have put them in the headings and in the tables so that a sentence lifted out of this page still travels with its label attached. If you quote us, quote the label.

The three labels, defined

One consequence of this rule is worth stating up front. The most-quoted numbers in this category — $11 billion, $14 billion, $2 billion — measure the microdrama format as a whole, not AI-produced titles. No authoritative dataset shows what percentage of any app's catalogue is AI-generated. Every market figure you see attached to the phrase "AI drama market" is therefore either a proxy or an error, and usually an error.

What do we actually know today?

Before forecasting 2027, it is worth being precise about the measured base, because a forecast inherits every weakness of the figures it starts from. These are reported outcomes rather than projections, and they are the floor everything else on this page is built on.

Two things stand out. First, the growth is real and measured, not projected — US monthly active users more than doubled in a single year and installs grew 238% year on year in Q1 2026. Second, every measurement of AI production specifically comes from China. The US has audience numbers and revenue numbers but no AI-share number at all, which means every US-facing statement about "how many AI dramas" is an inference from Chinese platform data.

The full sourced set, with methodology notes on each figure, lives in our AI drama statistics page. If you want the definitional groundwork first, start with what AI dramas are.

What do the published forecasts actually say?

Four published forecasts touch 2026 or later, from three research firms, and they use three incompatible definitions. This is the complete set as far as we can establish. If a piece of coverage quotes a forward number that is not in this table, it is either derived, misattributed or invented.

Why these numbers must not be combined

Omdia measures all microdrama revenue. Deloitte measures in-app purchases only. QYResearch measures AI streaming platforms on its own definition. Adding, averaging or ranking them against each other produces a number that describes nothing.

The gap is not small. Omdia's above-$14 billion for 2026 and Deloitte's $7.8 billion for 2026 are roughly a factor of two apart, and that gap is mostly definitional rather than a disagreement about the world: advertising revenue, web payments and third-party Android stores sit inside one scope and outside the other. Sensor Tower's underlying app-store estimates, which feed several of these, explicitly cover App Store and Google Play in-app purchases only.

The QYResearch line is the only forward number in existence that tries to isolate AI, and it does so by measuring platforms rather than titles. A platform with a mixed catalogue counts in full or not at all depending on how it is classified. We report the figure because it exists and because we would rather show you the whole set; we do not treat it as the size of the AI drama market, because nothing measures that.

How much should you trust a forecast?

A forecast is a research firm's paid opinion about the future, published on a date, using a method it usually does not fully disclose. It is not an audited outcome, it is not a guarantee, and in this category it has already been wrong in the optimistic direction and in the pessimistic direction within eighteen months.

The pessimistic misses are the more interesting ones. Global short-drama app revenue went from $178 million in Q1 2024 to nearly $700 million in Q1 2025 — roughly four times in a year, per Sensor Tower via Deloitte. Global downloads reached 2.3 billion in 2025 with Q4 download growth of 186% year on year, and Q1 2026 install growth ran at 238%. Growth at that rate outran what most published forecasting anticipated when the category was first sized. A firm that projected sensible double-digit growth in 2024 was badly wrong by 2025, in the direction nobody complains about.

That cuts both ways for 2027. A category that has repeatedly exceeded its forecasts can keep doing so; it can also revert hard, because the same conditions that produce four-times growth — cheap production, aggressive paid acquisition, a novel format — are the conditions that produce saturation. Omdia's ~$2 billion for US 2027 implies roughly a third of growth on the ~$1.5 billion 2026 forecast, a marked slowdown from the doubling seen between 2024 and 2025. Whether that deceleration is a considered judgement or a conservative default, Omdia has not published.

Three questions to ask of any forecast in this category

1. What is the scope? All revenue, in-app only, or one firm's segment definition. Without this the number is meaningless.
2. Does it isolate AI? Almost none of them do. If a headline says "AI drama market" and the underlying figure is Omdia's, the headline is wrong.
3. What was the same firm saying two years ago? Forecast revisions tell you more about a firm's method than the current forecast does.

We hold ourselves to the same three questions in the our calls section below, which is why those calls are deliberately narrow.

Will AI video look consistent by 2027?

TREND (evidenced) · Confidence: high. Spatiotemporal consistency in generated video is the acknowledged open research problem, and it will still be an open problem at the end of 2027.

The evidence is the literature itself. An ACM survey, A Survey: Spatiotemporal Consistency in Video Generation, published May 18, 2026, frames consistency as the central unresolved question in the field. OpenReview work on narrative and temporal consistency in long video, March 26, 2026, reaches the same conclusion for long sequences, and arXiv:2510.04999, Bridging Text and Video Generation: A Survey, October 6, 2025, describes the same gap. When a field publishes surveys calling something the open problem, that is a statement that it is not close to closed.

The production evidence agrees. Shots rarely exceed 10 to 12 seconds — a model constraint, not a stylistic choice. ByteDance's Jimeng runs to about 12 seconds per generation, or 30 with motion templates; Kling reaches roughly three minutes. A 90-second episode is therefore assembled from a dozen or more separate generations, and every seam between them is a place where a face, a costume or a prop can change. Seedance 2.0 in February 2026 was a real step up in quality and did not change that architecture.

Why the confidence is high rather than certain. Research timelines are unpredictable and a capability jump can arrive without warning. What we are confident about is narrower: the specific failure the format needs solved is not generation quality but identity persistence across a 40-to-100 episode season, and nothing in the published work is aimed at that horizon. We treat a breakthrough as possible and list it under what would surprise us.

Will apps ever label their AI titles?

TREND (evidenced) · Confidence: high for the current state, low for the direction. Not one of the thirteen apps we track labels individual titles as AI-generated, and there is no US mechanism requiring it.

This is an observed fact rather than an inference: we install every app on a US account and look. DramaBox, ReelShort, ShortMax, FlexTV, StoReel, Character.AI Series, DramaWave, GoodShort, VibeShort, FlickReels, MoboDrama, MyMuse and NetShort all present AI-produced and conventionally produced titles in the same interface with the same metadata. MoboDrama positions itself as a fully AI catalogue, which is a marketing claim about the whole library, not a per-title label — and not a verified fact.

The consequence is structural. Because no app labels, no dataset can be built, so nobody can measure the AI share of any catalogue, so every market figure attached to "AI dramas" has to borrow a microdrama figure instead. The absence of labelling is the root cause of the measurement problem this entire page has to work around.

The provenance technology exists and is partly deployed. Google's Veo 3.1 writes a SynthID watermark into its output. C2PA is a functioning provenance standard that records an asset's origin and edits — though it is neither a rights registration nor a detector, and it only helps if the tools in the chain preserve it. The gap is not capability. It is that no app has any commercial reason to be the first to tell US viewers which of its titles were machine-made.

Practical detection, in the absence of labels, comes down to watching for the seams. We keep a working method in how to tell if a drama is AI-generated.

Is interactivity the only genuinely new format?

TREND (evidenced) · Confidence: medium. Character.AI's (c.ai) Series, launched July 2026, is the only structural format innovation in the category, and format innovation is where the next differentiation comes from.

Everything else in this market is a variation on the same object: 9:16 vertical, 60 to 120 seconds an episode, 40 to 100+ episodes a season, hook in the first three seconds, cliffhanger at the end, paywall at roughly episode five. Thirteen apps compete on catalogue and price within an identical format. Character.AI Series changes the object itself — you watch an episode, then chat with the character in it. Core viewing is free, a c.ai+ subscription buys extras, and some features are gated to 18+.

StoReel's Canvas tool points the same way from a different angle, letting viewers build episodes rather than only consume them. One reported case put $150 against four one-minute episodes, which tells you the unit economics of viewer-side creation are not yet obviously good.

Why medium and not high. One launch is one launch. Character.AI Series has a short catalogue, and a novel interaction can fail on retention as easily as it can succeed. What supports the trend label is the underlying logic rather than the sample size: when thirteen products share a format, the cheapest available differentiation is to change the format, and generative systems make conversational characters nearly free to add. Our hands-on notes are in the Character.AI Series review.

Trend: volume keeps rising, the hit rate does not

TREND (evidenced) · Confidence: high. Cheap production has not made hits more common. It has made misses cheaper, and that arithmetic pushes the market toward consolidation rather than a long tail of successful independents.

DataEye counted roughly 221,900 AI dramas published on Douyin in the first half of 2026, about 1,200 a day, up from around 470 a day in January 2026. Of that half-year cohort, about 1,055 titles — 0.47% — passed 100 million views. In April 2026 Douyin carried around 44,200 AI titles against 3,248 live-action ones. This is Chinese platform data and the only place in the world where AI-drama output is counted at scale.

Put that next to the revenue split. A 50/50 platform-creator split is common, and a title reaching a billion views can return $100,000 to $500,000 to its creator. Those are real outcomes with roughly one-in-two-hundred odds attached. A studio that can fund two hundred attempts is playing a different game from a creator who can fund three, and the more the cost per attempt falls, the more decisive raw attempt-count becomes.

The US market structure points the same way for a different reason. Thirteen apps compete here, and nine of them give us no usable US price at all at all. A market where two thirds of the competitors will not quote a price is not a stable equilibrium; it is a market where nobody has yet had to compete on price because acquisition, not retention, is still doing the work. We score every one of them on billing transparency for exactly this reason — the method is on our how we test page.

What consolidation would look like in practice

GUESS (ours) · Low confidence. If consolidation happens it shows up first as catalogue mergers and app shutdowns among the smallest players — the apps with thin libraries, minimal documentation and no published pricing. FlickReels, MoboDrama, MyMuse and NetShort are the four we would watch, purely because they are the smallest and least documented, not because we have any information about their finances. We have none, and we are labelling this a guess for that reason.

Trend: acquisition eats the production saving

TREND (evidenced) · Confidence: high. Paid user acquisition is now often the largest cost line in an AI drama's budget, which means the headline story of collapsing production costs describes a shrinking share of the total.

The production side of the story is well documented. FlexTV vice-president Tang Tang told MIT Technology Review on May 15, 2026 that moving North American production to AI cut costs 80 to 90 percent and compressed schedules from three or four months to under one, with teams of about ten people — a vendor describing its own results, which is how we label it. A 60 to 80 episode AI microdrama has been produced for $3,000 to $14,000 (NOW News, December 15, 2025). Independent creator Jiang Lan accounted for roughly $280 per finished minute, about $28,000 for twelve eight-minute episodes (TVBS, August 7, 2026). Against a live-action vertical drama at roughly $2,500–$3,300 per finished minute, the saving is not in doubt.

The distribution side moves the opposite way. Q1 2026 installs grew 238% year on year, and an Adjust survey published June 23, 2026 found 67% of marketers advertising in or interested in short-drama apps. More bidders and more inventory in the same auctions means a rising price for the same install. APAC revenue per monthly active user ran at $1.45 in Q1 2026, which sets a hard ceiling on what an install can be worth in that region.

The cost lines a headline production figure hides are the ones that decide the outcome: failed generations, retakes, upscaling, voice correction, human editing, music and SFX, storage, translation and QC, distribution fees, legal review, refund handling — and paid acquisition on top of all of it. The metric that matters in production is cost per accepted shot, not cost per generation, because a 90-second episode consumes many minutes of raw generated material. The metric that matters commercially is cost per paying viewer, and nothing in the generative stack reduces it.

Where this ends up in 2027

GUESS (ours) · Medium confidence. We expect the conversation to shift from cost per minute to cost per paying viewer during 2027, because the first number has already been optimised to the floor and the second has not been optimised at all. There is no benchmark for it — no US figure exists for the cost of a completed, edited, legally cleared AI-drama episode, let alone for what it costs to find someone who will pay for it.

Trend: localisation is the real distribution engine

TREND (evidenced) · Confidence: high. AI localisation — dubbing, subtitling and cultural adaptation — is what puts a Chinese-produced title on a US phone, and it is the most commercially consequential use of AI in this category that almost nobody counts as AI production.

Look at what US viewers are actually served. DramaBox carries thousands of series with roughly 200 new titles a month across 200+ countries and extensive dubs and subtitles including Spanish and Hindi. ShortMax ships polished subtitles and dubs in many languages. A large share of that catalogue is conventionally filmed material reaching an English-speaking audience through a voice model. The production was human; the thing that made it reach you was not.

This matters for every number on this page. If AI-localised titles count as AI dramas, the AI share of US viewing is very large. If only AI-generated footage counts, it is much smaller and unmeasured. The definitional choice changes the answer by an order of magnitude, and no research firm states which convention it is using. We set out the five production types, including AI localisation, in what are AI dramas.

The direction is not seriously in question. Localisation is the cheapest possible form of catalogue expansion: no new story, no new shoot, no new rights beyond the territory, and an addressable audience that multiplies. US microdrama monthly actives went from 26 million to 66 million in a year on the back of catalogues largely built elsewhere.

Will the US regulate AI dramas?

TREND (evidenced) · Confidence: high for the described facts, none for US extrapolation. China has built a working AI-content regime for this exact format. The United States has not built anything specific to it. Both statements are facts about August 2026, and neither predicts 2027.

What China demonstrates is feasibility. A state that decides platforms must label AI content can make them do it, at scale, across the highest-volume AI drama market in the world. That is genuine evidence, and it is evidence about capability only. It says nothing whatsoever about US intent, and anyone using it to predict US policy is guessing with a Chinese prop.

The US surface for a producer is wide and mostly indirect: copyright, right of publicity, trademark, contract, music licensing, voice rights, labour agreements, false endorsement, privacy and platform disclosure rules, with state law differing. YouTube's monetisation policy is a live example of platform rules doing regulatory work — mass-produced, repetitive, low-value content may be ineligible for monetisation, while AI assistance alone is not an automatic bar where the content adds original value. Our full breakdown sits in are AI dramas legal in the US?

Are traditional studios using AI too?

TREND (evidenced) · Confidence: high. The interesting movement in 2027 is not fully synthetic microdrama. It is conventional production quietly absorbing generative tools, which is already happening in named shows at named companies.

Four of those six are not microdramas at all, and that is the point. The Netflix and Amazon examples mark the moment AI production stopped being disreputable in prestige television, which changes the labour and licensing conversation far more than another thousand Douyin uploads does. Arte's Paradoxes and Story TV's The Bose Network sit in the middle of the range at roughly half-AI workflows, which is where we would expect most serious production to land.

The labour evidence runs underneath all of it and is not comfortable. Reporting from the Chinese market describes microdrama pay falling by half or more, with some studios shrinking by around 70%. That is the volume tier, not the premium tier, and it is the clearest measured consequence of the cost curve in this page's acquisition section. We follow the argument in will AI replace actors.

What would surprise us

A forecast page that cannot say what would falsify it is marketing. These are the outcomes we consider unlikely enough that they would force us to rewrite several sections. Each is labelled with our own confidence, and all of them are guesses about probability, not published forecasts.

What nobody can forecast

Four questions decide most of what 2027 looks like, and no published forecast, trend line or dataset gives you leverage on any of them. Anyone who tells you otherwise is selling a guess with a chart attached.

Whether US regulators act

There is no US federal law specific to AI dramas as of August 19, 2026. The Copyright Office's Part 3 report on training data is pre-publication rather than a final rule. The FTC's impersonation rule covers government and business impersonation but its extension to private individuals is unfinalised. Each of those could complete in 2027 or sit unchanged. Regulatory timing is not a trend; it is a decision by a small number of people, and decisions of that kind do not have base rates worth quoting.

Whether a consistency breakthrough lands

Research does not arrive on a schedule. Seedance 2.0 in February 2026 was a real step up that nobody had dated in advance. The honest position is that the current literature treats consistency as unsolved and gives no indication of when that changes, and that a surprise is possible in any given quarter. A page that assigned this a percentage would be inventing information.

Whether audiences turn

All the reception evidence in existence is qualitative and it contradicts itself. Director Qingge Gao said in August 2026 that microdrama viewers "can easily watch AI minidramas… most of them don't care." Reviewer Jenny Cooper watched 25 AI titles and finished 3 in June 2026, concluding story remained decisive. Reddit threads return "This is AI slop. Yikes." Holywater's Bogdan Nesvit has said engagement metrics for AI content are in some cases comparable to live-action. Four real observations about four different audiences, and not one survey among them. No representative US complaint dataset exists either. We lay out both sides in AI slop: do viewers actually care?

Whether any platform starts labelling

This is the one that unlocks everything else. If one app labels, the measurement problem starts to dissolve, comparison becomes possible, viewers who care can filter, and the research firms get a denominator. If none does, 2027 ends exactly where 2026 ended: a category discussed in billions and measured in nothing. We have no evidence either way, and we are not going to pretend the absence of evidence is evidence of stasis.

What do we think will happen in 2027?

GUESS (ours). These are the editorial team's judgements, not sourced facts, not published forecasts and not derived from any dataset we have not shown you. We are publishing them because refusing to answer the question this page is named after would be its own evasion. Each one is written to be checkable on a date, and each carries a confidence level that means what it says.

Note what is not on that list. We have not predicted a 2027 revenue number, because we have no method for producing one and Omdia does. We have not predicted which app wins, because that is unknowable and would conveniently flatter whichever app we earn from. We do earn a commission when readers install DramaBox through this site, and DramaBox does not appear in our calls for exactly that reason.

How can you check whether we got it right?

Come back and hold this page to account. That is not a rhetorical flourish; it is the only thing that makes a forecast page worth writing. This version is dated August 19, 2026 and every call above carries December 31, 2027 as its check date.

What we commit to

• We date the page. The update stamp at the top is the date the figures were last verified, not the date the CMS touched the file.
• We revise it rather than replace it. When a published forecast is superseded, the old number stays in the table with its original date so you can see the revision.
• We mark our own calls wrong when they are wrong. A prediction that quietly disappears is worse than a prediction that was never made.
• We keep the labels. If a guess becomes a trend because evidence arrives, the label changes and we say when it changed.
• We name the gaps. The declared gaps on this page — no AI-share dataset, no US episode cost benchmark, no quality benchmark, no representative complaint dataset, nine of thirteen apps with no published US price — are findings, not omissions.

How to check us yourself

Every published forecast in the table above names its firm and its date, so you can go to the source rather than trusting our summary. Every trend section names the paper, the platform count or the production behind it. Our testing protocol — install on a US account, watch until the free tier runs out, pay through at least one paywall, keep the receipt — is written out on how we test, and the resulting scores for all thirteen apps are in the reviews hub. If you find a figure here that is wrong, tell us and we will correct it with a note saying what changed.

Frequently asked questions

The bottom line

For 2027 there is one published number, seven trends with real evidence behind them, and four questions nobody can answer. The published number is Omdia's roughly $2 billion of US microdrama revenue, reported via Reuters in August 2026, and it measures the whole vertical short-drama format rather than AI-produced titles. Nothing published anywhere sizes the AI drama market for 2027, because nothing measures the AI share of any catalogue.

The trends are the useful part. Consistency stays unsolved; no app labels; interactivity is the only new format; volume rises while the hit rate stays near half a percent; acquisition costs absorb the production saving; localisation is what actually moves titles across borders; China has a labelling regime and the US has nothing; and hybrid production is already normal in prestige television. Each of those is checkable against a dated source, and each could stop being true.

The honest ending is that the four unforecastable questions matter more than everything above them. If US regulators act, if a consistency breakthrough lands, if audiences turn, or if one platform starts labelling, most of this page needs rewriting. We would rather tell you that than sell you a chart. Come back on December 31, 2027 and mark our work.

DramaBoxEditor's pick — biggest short-drama library, lowest price to go unlimited 9.8Excellent

• Thousands of series, roughly 200 new titles every month
• Free episodes daily, plus ad-unlocks — no card needed to start
• Unlimited plans from about $5.99 a week, the lowest entry price we found
• Runs on Android, iOS and in a desktop browser

Go to DramaBoxAffiliate link · free to download · in-app purchases available

Affiliate link. We may earn a commission if you install through it, at no extra cost to you — it never changes our scores or placements. How we test · Disclaimer

Read next

Sources

• Omdia — global microdrama revenue of $11 billion for 2025 (October 21, 2025); above $14 billion forecast for 2026 via Broadband TV News (May 29, 2026); US revenue of ~$1.5 billion forecast for 2026 and ~$2 billion for 2027 via Reuters (August 2026); US monthly active users of 26 million (2024) rising to 66 million (2025); microdramas nearly twice the size of the global FAST market (March 2026). All scopes cover the whole microdrama format, not AI titles alone.
• Deloitte, 2026 TMT Predictions (November 18, 2025) — global in-app micro-series revenue of $3.8 billion in 2025 rising to a forecast $7.8 billion in 2026, in-app purchases only; ~30% of US Gen Z and millennials familiar with microdramas as of March 2025; China microdrama users of ~662 million in 2024, citing Chinese media.
• QYResearch (April 16, 2026) — AI short-drama streaming platform market of $1,006 million in 2025 rising to $2,301 million by 2032 on a 12% CAGR. One firm's segment definition, not a consensus figure.
• Fieldfisher (June 1, 2026) — US microdrama revenue of $1.3 billion for 2025.
• Sensor Tower — global short-drama app revenue of $178 million in Q1 2024 rising to nearly $700 million in Q1 2025, via Deloitte; US app revenue of ~$350 million in Q1 2025, about 49% of the global total (November 25, 2025); 2.3 billion global downloads in 2025, Q4 2025 download growth of 186% year on year and Q1 2026 install growth of 238%, via Adjust (June 23, 2026); Q4 2025 US daily engagement of 35.7 minutes per user for ReelShort against 26.9 for Prime Video, 24.8 for Netflix and 23 for Disney+, via Media Play News (April 27, 2026) and eMarketer (March 5, 2026). Sensor Tower's method estimates App Store and Google Play in-app purchases only and excludes advertising, third-party Android stores and direct web payments. The engagement figure is an intensity measure per app user, not proof of total viewing time.
• Adjust (June 23, 2026) — APAC revenue per monthly active user of $1.45 in Q1 2026; survey finding 67% of marketers advertising in or interested in short-drama apps in 2026.
• ACM, A Survey: Spatiotemporal Consistency in Video Generation (May 18, 2026); OpenReview on narrative and temporal consistency in long video (March 26, 2026); arXiv:2510.04999, Bridging Text and Video Generation: A Survey (October 6, 2025). The basis for treating consistency as the acknowledged open problem.
• DataEye — ~221,900 AI dramas published on Douyin in H1 2026, roughly 1,200 a day, of which ~1,055 (0.47%) passed 100 million views; ~44,200 AI titles against 3,248 live-action on Douyin in April 2026; ~470 AI titles a day in January 2026 via MIT Technology Review (May 15, 2026). China Netcasting Services Association via Global Times — more than 95% of new Chinese microdramas AI-generated in Q1 2026.
• MIT Technology Review (May 15, 2026) — FlexTV vice-president Tang Tang on an 80–90% production cost reduction, schedules falling from three or four months to under one, and teams of about ten people. A vendor describing its own results. NOW News (December 15, 2025) — 60–80 episode AI microdramas produced for $3,000–$14,000. TVBS (August 7, 2026) — independent creator Jiang Lan at ~$280 per finished minute, ~$28,000 for twelve eight-minute episodes.
• US Copyright Office — Copyright and Artificial Intelligence, Part 2: Copyrightability (January 29, 2025); Part 3 on training data remains pre-publication. SAG-AFTRA digital replica materials (November 17, 2025) and July 2025 interactive-media protections. Tennessee ELVIS Act in force July 1, 2024. California AB 1836 and AB 2602 (2024). FTC impersonation rule in force April 1, 2024 for government and business impersonation, extension to private individuals not finalised.
• China regulatory context — mandatory AI-content labelling from March 7, 2025; NRTA registration for AI microdramas budgeted at 800,000 yuan (~$110,000) or above; campaign against "AI magical re-editing" from January 1, 2026. iiMedia Research via CGTN (February 25, 2026) — Chinese AI manju market of ~18.98 billion yuan for 2025.
• Named hybrid productions — Netflix, The Eternaut, generative AI confirmed July 2025, building-collapse sequence roughly 10x faster than conventional VFX; Variety (March 20, 2025) on AI-assisted VFX in Amazon's House of David, episode 6; Arte France's six-episode Paradoxes; Story TV's The Bose Network, ~50% AI workflow, August 2026; CJ ENM's announced AI dramas and released AI animation; afaqs (September 9, 2025) on Dashverse's Raftaar.
• Reception evidence, all qualitative — director Qingge Gao via PetaPixel and Today (August 2026); reviewer Jenny Cooper, Vertical Drama Love (June 24, 2026), 25 AI titles watched and 3 finished; Bogdan Nesvit of Holywater via TheWrap (2026); a former Meta executive (April 2026) on a parallel category rather than a substitute; Business Insider (January 28, 2026) on ReelShort spending, with ~30% of surveyed users paying $25–$100 a month and ~27% paying nothing. None of these is a representative survey and none is presented here as data.
• App-level figures — DramaBox 90m+ registered users across 200+ countries (Investing.com, July 20, 2026) and Q1 2025 in-app revenue of $120m, +29% QoQ (Sensor Tower); ReelShort 49.9m MAU and ~27% share (November 2025) with Q1 2025 in-app revenue of $130m, +31% QoQ; StoReel's $34 million raise and 100-titles-a-month target (36Kr, March 26, 2026); Character.AI (c.ai) Series launch, July 2026; VibeShort 3.95 million downloads in 30 days, +346.58% year on year (May 2026); DramaWave 2.44x and FlickReels 3.75x Q1 2025 QoQ growth (Sensor Tower).

Affiliate disclosure. Some outbound links on this page, including the DramaBox links, are affiliate links. If you install through one we may earn a commission at no additional cost to you. It does not change what we write, and it is the reason DramaBox is deliberately absent from our own predictions above. Forecast notice. The forward numbers on this page are published forecasts by named research firms, not audited outcomes. They use incompatible scopes and must not be combined. The sections labelled GUESS are the editorial team's judgement and carry no data behind them beyond what is cited. Accuracy notice. Several figures here come from single sources or from vendors describing their own results, and we label those as such. Verified August 19, 2026; next scheduled review of our 2027 calls, December 31, 2027.