What AI Has Actually Taken From the VFX Pipeline

The question "will AI take VFX jobs" is the wrong shape. Some VFX work is discrete, repeatable and objectively checkable, and some of it is a judgement call about whether a shot looks right. Automation has always eaten the first category and never touched the second. What is new in 2026 is how much of the first category there turns out to be, and where in the world it sits.

Which VFX roles are actually exposed?

Three, according to the clearest published analysis: rotoscoping, paint and matchmove.

HDRI Consulting's three-year outlook names these as the jobs most likely to be eliminated by 2027, and gives the reason plainly: they are tasks a model is good at because they involve identifying complex patterns, and because the correct answer is "relatively discrete and non-subjective." A roto matte is either accurate to the edge or it is not. There is no house style in a matchmove.

The same analysis lists the roles it expects to be accelerated rather than removed:

|---|---|

The dividing line is not difficulty. Cloth simulation is harder than roto by any technical measure. The line is whether the work terminates in a verifiable answer or in a taste decision, and every role in the right-hand column ends in taste.

Why is this a geography problem more than a technology one?

Because the exposed roles are not evenly distributed across the world, and never have been.

HDRI's assessment is explicit that the jobs most likely to be eliminated are concentrated in countries with lower labour costs — India above all — while North America, the UK and Europe are expected to absorb productivity gains as more or better work rather than as fewer people. That is a statement about where a cost saving gets taken, not about where the software runs.

The scale involved is easy to underestimate. VFX Voice reports that Mumbai is the largest visual effects hub in the world by headcount, and that six of the ten fastest-growing VFX and animation hubs globally are in India. Between 85% and 90% of Indian VFX studio revenue comes from international work; domestic clients account for only 10–15%. Individual studio headcounts tell the same story — BOT VFX went from six people in 2008 to 745 in 2024, FutureWorks employs over 500, ILM's Mumbai facility around 400.

An industry that grew by being the place the pattern-recognition work was sent is, by construction, the industry most exposed when pattern recognition gets automated.

Is AI the actual risk, though?

Not on its own. The most recent large shock to Indian VFX employment had nothing to do with AI: Technicolor's collapse put roughly 2,000 workers out of a job.

That matters for how you read the forecast. A pipeline built on outsourced volume for a handful of international clients is fragile whether the disruption arrives as a model or as a bankruptcy. The failure mode is the same — a dependency on being the cheapest supplier of a commoditised task — and AI is one more way for that dependency to be called in.

The countervailing pressure is real too. Some clients respond to a productivity gain by buying the same work for less; others buy better work for the same money. VFX Voice quotes the cost argument still holding: "pound for pound, if you give me a sequence to bid and work on, India still comes out ahead." Both things can be true. The industry keeps growing and the entry-level rung of it keeps getting shorter.

What does this change for someone working in VFX now?

The honest advice is uncomfortable, because the roles being automated are the roles the industry has always used as its front door. Roto and paint are how people get in. Removing them does not just cost the people currently doing them — it removes the training path to everything in the right-hand column.

Three things follow.

Move toward judgement, deliberately and early. Compositing is the nearest rung and the one the analysis expects to survive. Look development, concept and rigging are further but safer. None of these are reachable by waiting.

Own something that requires context. A model can pull a matte. It cannot sit in a review, understand why the director keeps saying the shot feels heavy, and know that the answer is a grade note rather than a comp note. That kind of judgement is built on set and in review sessions, which is an argument for being in the room even when the work is remote.

Stop being findable only through one vendor. If your entire professional identity lives inside one studio's client list, a Technicolor-shaped event ends it. A public, verifiable body of work — real credits, real shots, attached to your name — is what survives a studio closing. That is the whole argument for a VFX artist's crew card existing independently of an employer, and it is the same argument as where crew actually get hired: the referral network is strong, but it only reaches people it can see.

Does this match what is happening to the rest of production?

Broadly, yes — with the same shape. Our read on what AI actually does in film production found automation landing hardest on discrete, verifiable, repetitive tasks and stalling on anything requiring physical presence or accountability. VFX is that pattern at its most concentrated, because more of the work is discrete and verifiable than in any other department.

It also rhymes with where the production jobs went: the losses are real, and they are relocations at least as often as they are disappearances. The work does not stop existing. It stops existing there.

For anyone running the VFX side of a production rather than working in it, the practical consequence is about tracking rather than headcount — knowing which shots are out, at which vendor, at which version, and what the review said. That is what a VFX pipeline tracker is for, and it becomes more important, not less, when the vendor mix is changing under you.

The honest caveat

The 2027 forecast is a forecast. HDRI's analysis carries no quantified headcount projection and no percentage of jobs at risk, and the author is explicit that client behaviour — cheaper versus better — could send the outcome either way. Treat the direction as well-argued and the timing as a guess.

If you are an Indian VFX artist reading this as a career question rather than an industry one, the adjacent practical piece is how freelance film crew in India should handle tax and invoicing. Independence has paperwork attached, and the time to learn it is before you need it.