A Glimpse of Generative AI in 2027

Most brands are building their 2027 plans this fall, and the generative AI line is the hardest one to write. Analysts, agencies and consultancies have all published views on next year. Some of what gets quoted as a 2027 prediction is in fact a legal deadline already on the calendar. The rest is forecast: an estimate by a named firm, made on a given date, that will be revised.

This piece keeps the two apart. Dates in law come first. Forecasts follow, each with its source and the month it was published, with a focus on what touches marketing, video and creative work.

Already on the calendar

Three dates in 2027 are written into law. On January 1, California’s AB 853 requires large online platforms to detect whether content they distribute carries provenance data that follows widely adopted standards. From the same day, a platform that hosts generative AI systems may not knowingly offer one that fails to place the disclosures required by the state’s AI Transparency Act.

In Europe, the AI Act sets 2 August 2027 as the date by which providers of general-purpose AI models already on the market before 2 August 2025 must meet the rules for those models. The same day, each member state must have at least one AI regulatory sandbox running. On 2 December 2027, obligations for the stand-alone high-risk systems listed in Annex III begin. That date moved: it was August 2026 until the Digital Omnibus on AI, Regulation (EU) 2026/1744, entered into force on 27 July 2026.

What did not move matters more to marketers. White & Case notes that the omnibus gave no extension to the Article 50 transparency rules, which have applied since 2 August 2026, and that generators already on the market before then have until 2 December 2026 to mark their output. So from the first day of 2027, generated video reaching European viewers should carry a machine-readable mark from the system that made it, and large platforms serving California must look for that kind of data.

Where the money is forecast to go

Gartner’s latest forecast, published in September 2026, puts worldwide AI spending at about $3.6 trillion in 2027, up 36% from 2026. Infrastructure takes the largest share, close to $2 trillion. Spending on generative AI models themselves is forecast to grow from $28.3 billion in 2026 to $51.6 billion in 2027.

Not every forecaster saw an open tap. In October 2025, Forrester predicted that enterprises would defer a quarter of their planned AI spend into 2027, because fewer than one-third of decision-makers could tie AI’s value to their company’s financial growth. McKinsey’s 2026 survey of 1,719 respondents shows where the payoff is felt: revenue gains are most often attributed to AI in marketing and sales.

Advertising is shifting with it. WPP Media’s June 2026 forecast calls generative search the fastest-scaling channel ever recorded, growing from $5.1 billion in ad revenue in 2026 to over $100 billion by 2030. In its 2027 predictions, released in September, Forrester expects conversational advertising inside AI assistants to become a multibillion-dollar channel next year. It also expects “recommendation share,” whether an AI system recommends a brand while a customer researches, to become a leading media measure.

Forecasts aimed at 2027 and just beyond

Each figure is an estimate by the firm named, dated to when it was published

Attributed forecasts for generative AI, 2027 to 2030

Source and date	Forecast	Horizon
Gartner, September 2026	Worldwide AI spending of about $3.6 trillion, up 36% on 2026	2027
Gartner, June 2025	Over 40% of agentic AI projects canceled	End of 2027
Gartner, November 2024	40% of existing AI data centers operationally constrained by power availability	2027
Gartner, January 2026	Brands allocate 50% of influencer marketing budgets to content and creator authenticity initiatives	2027
Forrester, September 2026	Conversational advertising becomes a multibillion-dollar channel	2027
Gartner, January 2026	60% of brands use agentic AI for one-to-one interactions	2028
WPP Media, June 2026	Generative search ad revenue above $100 billion, from $5.1 billion in 2026	2030

Read the board as a set of bets. Forecasts get revised: WPP Media raised its 2026 global ad growth estimate from 7.1% in December 2025 to 8.9% six months later. A number published a year before 2027 begins describes the forecaster’s assumptions on that day.

Agents draw the boldest forecasts and the most cautious

Agentic marketing, where AI systems plan and carry out tasks across channels with little human input, pulls forecasters in both directions. In January 2026, Gartner predicted that by 2028, 60% of brands will use agentic AI for streamlined one-to-one interactions, with agents acting as “persistent digital concierges” across marketing, sales and support.

The same firm had warned in June 2025 that over 40% of agentic AI projects will be canceled by the end of 2027, citing escalating costs, unclear business value and inadequate risk controls. Gartner also estimated that only about 130 of the thousands of vendors selling agentic AI offer the real thing, with the rest relabelling existing products. Anushree Verma, a senior director analyst, described most current projects as “early stage experiments or proof of concepts.”

Actual use sits between those poles. In McKinsey’s 2026 survey, 40% of respondents from organizations with more than $1 billion in revenue report scaling AI agents, up from 27% a year earlier; among smaller organizations the share stayed at 22%. Both forecasts point to the same practice: give agents bounded jobs with a measure attached, such as producing and checking format variants of an approved video, and judge them on cost per usable asset.

Cheaper per answer, costlier per watt

Compute pulls in two directions. In March 2025, Epoch AI measured that the price of reaching a fixed level of performance on language model benchmarks had been falling between 9 and 900 times per year, depending on the task. The researchers cautioned that the fastest drops were recent and may not persist.

Power is the counterweight. In November 2024, Gartner predicted that 40% of existing AI data centers will be operationally constrained by power availability by 2027, and that rising power costs will be passed on to providers of AI products and services. Its September 2026 forecast describes demand for AI infrastructure as “strong and inelastic to pricing pressures.” McKinsey found that one in five organizations is already limiting AI use because of operating costs.

Generated video is among the most compute-hungry uses of AI. A brand should not assume that the price of a generated second will follow text prices down the same curve, and should expect vendors to change pricing or retire models during a year-long contract.

Trust turns into a budget line

Gartner predicts that by 2027, brands will put half of their influencer marketing budgets into content and creator authenticity initiatives, including identity verification, provenance checks and anti-deepfake measures. In its consumer survey, 78% of respondents rated explicit labelling of AI-generated content as very important, or as the most important factor, in keeping their trust.

Set that forecast beside the fixed dates and the direction holds even if the percentage misses. Provenance data attached at generation, kept through the edit and read by platforms is becoming the normal path for a generated asset. A production workflow that strips that data on export works against both the law and the audience.

What to prepare, whichever forecast is right

• Budget AI lines as ranges, each tied to one business metric, so a cut or a deferral does not stall the whole plan.
• Ask every video supplier how generated assets are marked, whether provenance data survives editing and export, and what disclosure text ships with each file for Europe and California.
• Write price-change and model-retirement terms into year-long contracts, and keep approved reference images, scripts and project files in your own storage so the work can move to another model or vendor.
• Pilot agents on bounded, measurable tasks, with a date to stop if the numbers don’t come.
• Start tracking whether AI assistants mention your brand when customers research your category.

Each of these pays off in a year when the forecasts hold, and in one where they miss.