You can't exit a feature

Edition 9 mapped the intermediation layer — the entities sitting between financed grid capacity and the commercial buyers who can't access it at scale. It closed by asking whether those entities are building intermediation as a standalone business, or as a defensive moat for an adjacent position they already hold.

This edition answers that. The answer changes what you're buying.

On 12 March 2026 the AEMC (Australian Energy Market Commission) released a draft rule proposing ride-through technical standards for large data centres connecting to the NEM (National Electricity Market). Eleven days later the federal government published formal expectations of data centre and AI infrastructure developers. The AEMC's framing was blunt: data centres are no longer passive loads. They're active grid participants — and when they fail to ride through a fault, the consequence can be cascading failure.

Read narrowly, that's a grid-stability story. Read for capital, it's something else. A regulator has begun pricing the quality of delivery, not merely the existence of a connection. That distinction — between holding a contract and actually delivering firm power against it — is the fault line that separates a standalone intermediation business from a defensive moat.

Is the layer a standalone business with its own cash flows, financing and exit — or a defensive capability that exists to protect generation, land, grid access, or a data centre campus that someone already owns?

The test is mechanical, not philosophical. For any intermediation asset, ask one question: could this business survive if the adjacent asset disappeared?

If the generation portfolio vanished, does the access-structuring platform still have value? If the data centre campus vanished, does the co-located power asset still have a market? If the land position vanished, does the intermediation layer still have economics?

A standalone business answers yes — its own counterparties, its own cash flows, its own market. A defensive moat answers no: its returns flow into the adjacent asset's P&L and can't be detached.

This maps onto the three-layer framework from Edition 7. Physical capacity (Layer 1) and commercial architecture (Layer 2) can be assembled by anyone with capital. What makes intermediation defensible — or captive — is Layer 3: who holds the access rights, the dispatch position, the regulatory standing. The moat builder uses Layer 3 to wall off an adjacent asset. The standalone operator turns Layer 3 into a product it can sell to anyone who pays.

The new Australian rule matters here for a precise reason. The rule targets loads, not aggregators. But it signals the regulator's direction: delivery quality is becoming a measured standard, not a contractual nicety. If that logic increasingly reaches supply-side contracting, the gap between a paper certificate and firm, time-matched power stops being pure margin and starts becoming a liability.

Across all four markets, a 2026 regulatory event is actively sorting standalone businesses from defensive moats: India's ISTS cutoff, Australia's shift toward more granular service definition, Vietnam's DPPA eligibility expansion, Malaysia's AI-prioritisation policy, the Philippines' open first-mover question. The sorting is happening now, in the structuring choices being made this year.

For the investor, the discipline is to stop underwriting the sector and start underwriting the separability. Two assets can look identical — same megawatts, same offtakers, same firming stack — and be worth radically different multiples depending on whether the cash flows can be detached and sold. The moat may earn well. But you can't exit a feature.

The honest caveat: a clean, completed sale of a standalone APAC intermediation platform to a financial buyer, on separable-cash-flow merits, hasn't yet printed. The structures are forming faster than the exits. That gap is the opportunity — and the risk.

If the intermediation layer is absorbed into the vertical stacks of the operators who need it most, who's left to serve the mid-market buyers that hyperscalers will never contract with directly — and what does the capital structure of that specialist lane look like once the moat builders have already captured the anchor tenants?

Edition 11 picks that up.