When approving the next AI budget, a factory needs to know who can stop a robot if a sensor misreads a part. Finance needs the cost of completing one task. And when a temporary app is retired, someone must revoke its access to customer records. All three decisions belong in the original investment proposal.

Gartner’s report of 30 October 2026 sets out ten global forecasts for 2028–2030 to inform planning in 2027. The figures below are Gartner’s forecasts. The practical examples are ALPHASEC’s recommendations for Thai organisations.

1. Physical AI joins frontline workers

By 2030, 80% of frontline workers at international companies will have assistance from physical AI, such as robots that sense and respond to their surroundings.

ALPHASEC’s view Pilot a single inspection station. Measure both the reduction in defective parts and how quickly a worker can safely stop the equipment after a sensor error.

2. Autonomous agents put public services under pressure

By the end of 2030, more than 10 billion agents created by individuals, businesses and governments will congest public services as they submit requests on users’ behalf.

ALPHASEC’s view Verify whom each agent represents and the authority it holds. Preserve access for people using the service directly.

When finance and energy teams share AI decisions

The same compute bill can reflect rising sales or unusual use. To find out what the money bought, service owners, finance, security and infrastructure teams need to compare their data.

3. AI spending becomes an attack target

By 2030, 80% of organisations offering public-facing AI will have experienced an attack intended to drive up their AI costs.

ALPHASEC’s view Before a public assistant goes live, have finance and security agree on a spending limit, an alert recipient and an alternative service for customers.

4. A dedicated team links AI to profit

By 2029, 60% of organisations deploying AI will have a dedicated function connecting total AI compute costs to business value and profit margins.

ALPHASEC’s view For an AI quotation tool, ask sales to measure time saved, contracts won and margin after AI costs before expanding it.

5. Costs are controlled as a model runs

By 2028, 60% of global enterprises will embed AI FinOps controls at inference, when a model processes a request.

ALPHASEC’s view Match model choice to task difficulty, cap retries and check answer quality when moving to a lower-cost model.

6. Enterprises help balance electricity grids

By 2030, more than US$1 trillion in enterprise energy assets will challenge utilities’ traditional role in grid balancing as companies earn revenue from capacity, surplus power and flexible demand.

ALPHASEC’s view At a factory or data centre, bring facilities and IT together to assess electricity demand, storage investment and the effect of an energy-control outage.

Who can explain what AI did, even after the system is retired?

7. The CIO or CAIO takes custody of AI evidence

By 2030, 80% of global enterprises will assign their chief information or chief AI officer responsibility for the evidence needed to demonstrate accountable AI use.

ALPHASEC’s view Sample one AI-assisted decision. Trace the data and model version, who approved the outcome and who could change it.

8. Insurance terms shape AI governance

By 2030, insurers will take a leading role in AI governance through stricter underwriting standards for AI liability cover that affect the cost of insurance.

ALPHASEC’s view Before renewal, ask which activities are covered, which exclusions apply and what test evidence the underwriter requires.

9. Modular AI offerings evolve faster

By 2029, 25% of global enterprises will continuously improve modular AI-powered offerings, making them harder for fast followers to copy.

ALPHASEC’s view Design product recommendations so they can change independently of checkout. Track sales, complaints and the ability to roll back each release.

10. Many new applications have a planned life under a year

By 2029, 80% of new applications will be designed for less than a year of use, for example a single campaign.

ALPHASEC’s view On approval, record a retirement date, a data owner, who revokes access and which records must remain after the campaign.

ALPHASEC’s proposed agenda for 2027

Choose one customer-facing or production AI project. Ask the business owner and CFO to review cost per successful outcome, the CIO or CISO to demonstrate who can pause and restore the service, and the risk team to trace one event. Gaps in that exercise identify investments with a named owner and a result that can be verified.

Gartner’s Top Strategic Predictions for 2027 and Beyond · 30 October 2026 · G00860626, p. 3