Executive summary

Use-case map cells by maturity tier
Indonesia cold chain · 8 process areas x 4 lenses = 32 cells · Oct 2026
Proven several live, results disclosed 2 vaccine and room logging Scaling live at leading operators 6 sensors, slotting, multi-drop routes Experimental pilot, MoU or demo only 11 No Indonesian case tools exist abroad, no result here 13 energy, pricing, halal tracking
Source: [Kira estimates] from the use-case map in chapter 2

The answer in three lines

Proof of temperature is live. The state vaccine chain runs on more than 6,000 temperature loggers [UNDP SMILE]; commercial stores log rooms automatically [MGM Bosco 2026], and trucks carry trackers where shippers pay for them [TechCrunch McEasy].

Prediction is not. We found no Indonesian case of machine-learning energy control, compressor maintenance or space pricing, while abroad one operator cut 33 mn kWh a year [DOE 2019].

The owner of the asset keeps most of the gain: routes and power are cost lines; proof is what shippers buy, and four foreign investors and a Korean lender are rebuilding the hubs [LNEWS Jul 2026; BusinessWire Aug 2026].

Value at stakeWe size AI at about USD 0.13-0.35 bn a year by 2030, 2.5-7% of cold chain services revenue [Kira estimates].
A Japanese or Korean firm should sell or install the temperature and route data layer now, and buy into hubs only where it controls the data.
BusinessWire Aug 2026 · DOE 2019 · LNEWS Jul 2026 · MGM Bosco 2026 · TechCrunch McEasy · UNDP SMILE · Kira estimates (full titles in the source index)

Temperature logging, records and multi-drop routing are both data-heavy and digitisable; driving and loading are neither

Exposure differs by activity, not by company: the same operator is highly exposed in the room and barely exposed at the dock.

Cold chain activities by how automatable they are
Indonesia · KIRA scores 1-5 · bubble = share of operating cost · Oct 2026
Automate now Data first Mechanise later Tools exist, small prize How far the work can be digitised (1-5) Information intensity (1-5) Temperature logging and alerts Records and documents Multi-drop route planning Inventory and slotting Refrigeration energy control Inter-island line haul Compressor maintenance Loading and picking Driving
Source: [Kira estimates]

Why the room leads and the road lags

The room is a fixed sensor site. A store has power, connectivity and one owner, so its temperature can be logged without anyone acting: MGM Bosco states that its stores run automatic digital temperature control [MGM Bosco 2026].

The road is not. A reefer truck changes hands, crosses islands and loses signal; only shippers who pay for a sensor see cargo temperature [TechCrunch McEasy].

Exposure is not the same as value. Driving and loading are a large cost but barely digitisable at today's wages: drivers take home Rp5-10 mn a month [Aptrindo Jul 2025].

Scoring ruleDigitisability = share of the task a system can do or check without a person; information intensity = how much the task depends on data.
Start where the work is already data and only the decision is manual: alerts, records and route plans, not forklifts or drivers.
Aptrindo Jul 2025 = Gridoto citing Aptrindo Central Java · MGM Bosco 2026 = MGM Bosco Logistics · TechCrunch McEasy = TechCrunch · Kira estimates = KIRA analyst triangulation