When Private Power Reaches the Kitchen: The Public-Service Guardrails MBG Needs

MBG Watch · 2026-10-14

The premise

Energy resilience is no longer only a public-utility question. Large private buyers — data centers, industrial users, cloud firms, and clean-power offtakers — are beginning to finance generation, transmission, and upgrade programs that can add capacity to wider grids.

That may be useful. It may also be misread.

For MBG, the accountability question is narrow: when a kitchen, cold room, water pump, route hub, battery, or local grid upgrade sits near privately financed energy capacity, what must be public before that capacity is counted as meal-service resilience?

The answer is not “private finance is unsafe.” The answer is also not “any new capacity nearby protects the kitchen.” The child-serving route needs a record of what it can actually rely on when heat, outage, price stress, maintenance failure, or dispute arrives.

MBG Watch has already argued this line from several directions: in “The Power Behind the Plate,” energy readiness is part of food safety; in “Not the Battery, the Continuity Record,” the asset matters less than the service continuity it can prove; in “Not the Virtual Power Plant, the Kitchen Flexibility Record,” flexibility is only accountable when the operating record is visible; in “The Cold Room Is a Budget Line,” cold-chain protection has to survive rupiah and energy stress; and in “Reuse Before Replacement,” existing infrastructure should be counted only after it is inspected, documented, and assigned.

This piece adds one boundary: not who funded the asset, but what the MBG service chain can prove when the asset is needed.

What the evidence supports

The current signal is real: private demand is becoming a financing force for power infrastructure.

Grist reported that Georgia regulators approved a structure under which Google will pay a subscription fee, on top of its electricity bill, to help cover upgrades at Georgia Power nuclear plants. The arrangement is expected to add about 96 MW of nuclear capacity, and Google receives credit toward its carbon-free energy goal. The same report notes that Google announced a separate partnership with Constellation Energy to fund nuclear upgrades in Illinois, New Jersey, and Pennsylvania, adding 890 MW to the PJM grid, while Amazon has announced another nuclear-capacity arrangement in Maryland.

The governance lesson is not that Indonesia should copy U.S. nuclear finance. It is that private offtake can make infrastructure appear publicly useful before the public-service terms are fully inspectable. Grist also notes that precise details of the Georgia Power–Google agreement were not public. That is exactly the point for MBG: an asset can add megawatts to a system while still leaving the kitchen-level reliance claim unclear.

Indonesia has its own, more relevant version of the signal.

Antara reported that Pertamina Geothermal Energy is positioning geothermal power to support green data centers, with internal projections that national data-center capacity could rise from 520 MW in 2025 to 1.8 GW by 2030. PGE framed geothermal as continuous baseload power suited to hyperscale users that require uninterrupted supply.

Ecobiz Asia reported a sharper grid-planning signal from PLN: potential Indonesian data-center electricity demand of around 13 GW, with 3,534.8 MW classified as Priority 1 and ready to be served, 9,460.8 MW still being processed, and contracted capacity projected at 15,944 MVA by 2028. PLN’s Fauzi Arubusman said that demand is no longer in the range of 100–200 MW; some orders are already 1 GW, 2 GW, and 3 GW. He also said that a 1 GW data center requires at least 500 kV support, that PLN normally needs three to five years to build 500 kV transmission, while data-center investors may expect two-year timelines.

A separate company announcement carried by TMCnet said BDx Data Centers had secured 788 MVA for its CGK4 AI campus in Jatiluhur, expanded CGK3A in South Jakarta to about 60 MVA, and advanced CGK5 in Suryacipta with up to 385 MVA of secured power capacity, bringing its Indonesian secured power portfolio to 1.2 GW in strategic partnership with PLN.

These are not MBG kitchen claims. They are infrastructure signals around the same national power system in which MBG kitchens, cold rooms, water systems, transport hubs, and local suppliers operate.

MBG’s own exposure is large enough that this boundary matters. Antara reported that Indonesia allocated Rp335 trillion for MBG in 2026, targeting about 82 million beneficiaries, with SPPG kitchens paid upfront and expected to stimulate local production and employment. At that scale, power reliability is not a background condition. It is part of whether food can be stored, cooked, cooled, transported, and served safely.

Food-safety evidence already points in that direction. Warta Ekonomi reported that 1,256 SPPG units in eastern Indonesia were temporarily suspended after inspections found technical and hygiene noncompliance, and that food-technology expert Yuyun Anwar urged cold-chain catering to keep temperature stable from production through distribution. The same report cited BGN’s emphasis on food-safety standards, wastewater treatment, hygiene certification, inspection, and laboratory testing before suspended units return to service.

Cold chain is not only a refrigerator. It is a power, maintenance, route, and record system.

What the evidence does not support

The evidence does not show that a specific private energy upgrade is currently protecting an MBG kitchen.

It also does not show that private infrastructure is currently endangering MBG.

The record supports a readiness standard, not an incident claim. It supports asking what MBG must publish before any public statement treats nearby, shared, privately financed, or private-demand-driven capacity as kitchen resilience.

There is another limit. U.S. nuclear uprate deals are not directly transferable to Indonesia’s MBG operating environment. Indonesia’s relevant signals include PLN transmission planning, geothermal baseload claims, data-center load growth, industrial estates, local diesel backup, batteries, cold-chain routes, and the practical reality of kitchens operating in 3T and other constrained regions. The comparator is governance, not technology.

The public record MBG needs

MBG should require a shared-infrastructure reliance record for every kitchen or route that counts externally financed, privately financed, or shared energy capacity as part of its resilience plan.

That record should be public enough for citizens, auditors, suppliers, and local governments to inspect the claim without seeing private commercial terms that genuinely need protection.

At minimum, the record should answer ten questions.

1. What load is actually protected?

MBG should list the kitchen, cold room, pump, water-treatment unit, route hub, charging point, freezer, communications node, or supplier facility covered by the resilience claim.

A nearby grid upgrade is not the same as protected kitchen load. A data-center feeder is not automatically a school-meal feeder. A renewable certificate is not cold-chain continuity.

2. What service level is promised?

The record should state voltage quality, interruption limits, backup duration, restart time, and the amount of MBG load that can be served during constraint.

“Connected to reliable power” is too vague. A cold room needs hours and temperatures. A kitchen needs operating windows. A distribution route needs departure-time certainty.

3. Who gets priority during shortage?

This is the core question. If a privately financed asset or grid upgrade supports several users, the record should say where MBG sits during outage, congestion, fuel shortage, maintenance, or curtailment.

If the answer is “ordinary customer priority,” say that. If child-serving kitchens receive emergency priority, show the instrument that grants it.

4. Who pays when conditions change?

A resilience claim has a cost path. MBG should disclose whether costs fall on the central budget, local government, kitchen operator, supplier, school, private offtaker, utility tariff, emergency fund, or later pass-through.

Rupiah Stability Watch has already warned that food safety in 3T areas becomes more fragile when heat, fuel, currency stress, and energy readiness interact. MBG’s own 2027 meal-tray ledger work has made the same point from the meal side: if electricity, diesel, cold chain, and procurement costs move, the tray record has to show where the shock lands.

5. Who maintains the asset?

A generator, battery, transformer, geothermal supply arrangement, feeder upgrade, cold room, or water pump can fail through maintenance neglect rather than design failure. MBG should publish who maintains it, how often, and which failed inspections trigger service suspension or fallback.

6. What is the fallback?

The record should name the fallback path: alternate kitchen, route rerouting, temporary cold storage, diesel backup, battery runtime, reduced menu, delayed service, or suspension.

The least honest resilience record is the one that only describes normal operations.

7. What notice does MBG receive before failure?

If the infrastructure provider knows about scheduled maintenance, grid constraint, curtailment, planned outage, fuel delivery risk, battery degradation, transformer overload, or price event, MBG should disclose the notice standard: who receives it, how quickly, and how the kitchen or route changes status.

8. What data is visible after failure?

After an outage or cold-chain break, MBG should publish time, duration, affected load, temperature excursion if relevant, meals held back, meals discarded, route changes, child-serving impact, and corrective action.

That is not punishment. It is how a public food program learns without hiding the cost of learning.

9. What commercial facts are protected, and what public facts are not?

Some private contracts will contain confidential pricing and negotiation terms. MBG does not need to publish every clause.

But confidentiality should not hide the public-service claim. Service priority, protected load, fallback, failure notice, safety correction, and cost pass-through are public-accountability facts when they are used to justify meal-service resilience.

10. Who can correct the record?

The record should name the office responsible for correction if a resilience claim proves false. A local operator, BGN, PLN unit, private partner, school, supplier, or district official should not be able to point in a circle while meals spoil or routes fail.

The least-harm path

MBG does not need to reject private-financed infrastructure. It needs to refuse uninspected resilience claims.

The least-harm path is practical.

First, reuse and disclose existing infrastructure before treating new capacity as proof. If a kitchen already depends on a PLN feeder, local generator, water pump, cold room, or supplier warehouse, inspect that chain first. “Reuse Before Replacement” remains the right discipline: count what exists only after the record shows condition, ownership, assignment, and failure path.

Second, separate capacity from continuity. A megawatt figure can describe a grid or a data center; it does not describe whether 3,500 meals stayed out of the temperature danger zone, whether a kitchen could cook through a morning outage, or whether a rural route had fuel and cold storage when the road changed.

Third, publish reliance by route, not by slogan. A national statement about energy readiness should be backed by route-level records: kitchen ID, protected assets, backup duration, cold-chain threshold, outage history, fallback route, cost exposure, and correction owner.

Fourth, make priority explicit before crisis. If child-serving kitchens are meant to receive special treatment during electricity constraint, that should be written before the outage. If they are not, MBG should not imply that they are protected.

Fifth, treat private-demand upgrades as possible public benefit, not automatic public capacity. A Google-funded nuclear uprate, a geothermal data-center partnership, a PLN transmission buildout, or a hyperscale power commitment may add capacity to a system. MBG can only count it when the kitchen’s claim is testable.

What I am uncertain about

Three uncertainties matter.

First, the publicly available record does not show how often MBG kitchens currently rely on shared private infrastructure, dedicated backup assets, or nearby industrial-grid capacity. That is exactly why the reliance record should exist.

Second, Indonesia’s data-center and green-power announcements are moving faster than the public evidence on local service priority. A project can be serious, financed, and useful while still saying little about food-service resilience for nearby communities.

Third, MBG’s internal operating records may already contain some of the facts named here. If so, the accountability gap is not necessarily absence of management; it is absence of public proof.

The line to hold

Private power can help a public meal program only when the public-service claim is visible.

For MBG, the question is not whether a data center, geothermal project, nuclear uprate, battery, industrial estate, or grid upgrade is good or bad in itself. The question is whether a child-serving kitchen can prove what happens at 6 a.m. when the power is constrained, the cold room is warming, the route is waiting, and the contract language is somewhere else.

Until that proof exists, private capacity nearby should be described as context — not counted as MBG resilience.