Around 219 facilities now sit inside the Safeguard Mechanism, and roughly 140 million tonnes of covered emissions pass through it each year. That’s the baseline against which every cement kiln, alumina refinery and LNG train in the scheme gets measured, and the number that matters isn’t the 140 million. It’s the 4.9 per cent a year it has to fall by, compounding, out to 2030 and beyond under the policy settings the Clean Energy Regulator administers.
Compounding is the word that gets lost in most coverage of this scheme.
A 4.9 per cent annual decline doesn’t feel dramatic in year one. By year five it’s a different plant. Heavy industry operators who signed up for the reformed scheme in 2023, when the Albanese government tightened the mechanism, are now three years into that curve and starting to hit the part where the baseline has genuinely diverged from what the facility can achieve through efficiency tweaks alone. Follow the money and you can see where this is heading: either operators buy Australian Carbon Credit Units, or they buy Safeguard Mechanism Credits off someone else’s overachievement, or they actually decarbonise the process. Most are doing some mix of the first two while promising the third.
What the scheme was built to do #
The Safeguard Mechanism covers facilities emitting more than 100,000 tonnes of CO2-equivalent a year, which in practice means the 200-odd biggest industrial emitters in the country: gas processing, steelmaking, alumina, cement, coal mining’s fugitive emissions, some data-centre-adjacent manufacturing. It doesn’t touch the grid directly. Origin and AGL’s coal and gas generators aren’t in it in the way you’d expect, because electricity generation has its own separate decarbonisation pathway through the market and through state schemes. This is an industrial-emissions tool, not a grid tool, and that distinction gets blurred constantly in the general commentary.
The 2023 reforms did three things that matter here. They set that declining baseline trajectory instead of the flatter one the Coalition had designed. They introduced international best-practice benchmarks that new facilities have to meet from day one, rather than inheriting a grandfathered baseline. And they created the tradeable credit, the Safeguard Mechanism Credit, so a facility that beats its baseline can sell the headroom to one that doesn’t. That last piece is the part doing the real work, and it’s also the part nobody outside the compliance teams at these companies properly understands.
Why 2026 is when it starts to bite #
Three years of 4.9 per cent compounding gets you to a baseline sitting somewhere around 14 to 15 per cent below where a facility started. That’s not trivial for a continuous-process plant. Alumina refining, LNG liquefaction, cement clinker production. These aren’t processes you tune at the margin and find 15 per cent. You find 2 or 3 per cent through better heat recovery and electrification of smaller loads, and then you’re staring at a gap that only capital investment or credits can close.
The Clean Energy Regulator’s own Safeguard data, published through its annual facility-level reporting, shows net Safeguard Mechanism Credit issuance has been running well short of what the scheme architects modelled. Fewer facilities are beating their baselines by wide margins than expected, which means the credit market is thinner than it needs to be for the mechanism to work as a genuine trading scheme rather than a slow-motion compliance cost. When supply is thin, the honest read is that credit prices firm up, and firming credit prices mean the cheapest compliance option increasingly looks like an Australian Carbon Credit Unit purchase rather than an actual process change. That’s not decarbonisation. That’s accounting.
Gas and LNG carry a disproportionate share #
Woodside, Santos and the other LNG exporters sit inside the Safeguard Mechanism with some of the largest single-facility baselines in the country, because liquefaction is energy-intensive by nature and because these plants run close to 100,000 tonnes a year almost as a rounding error. Santos has leaned on the Moomba carbon capture and storage project as its primary compliance lever, a point worth watching given how central that project now is to the company’s broader east-coast gas position, something I’ve covered in more detail looking at Santos’s position at the centre of the east-coast gas standoff. Woodside’s Pluto and Scarborough expansion, meanwhile, add fresh Safeguard-covered capacity at a moment when the baseline trajectory is tightening, not loosening, a tension explored in the company’s own position after Scarborough.
Here’s the bit the gas industry’s critics and defenders both underplay: a declining baseline on an expanding asset base means the compliance burden isn’t flat, it’s actively getting steeper on two axes at once. New capacity comes in in at a tighter international benchmark than existing capacity did when it was built. That’s by design. It’s also why the LNG sector’s lobbying on Safeguard settings has been some of the most consistent pressure the scheme has faced since the 2023 reform passed.
The trade-exposure argument, and why I don’t fully buy it #
The standard defence from heavy industry is that Australia can’t ask its cement and steel and alumina producers to absorb a carbon cost that competitors in Indonesia, China or India don’t face, because the product is tradeable and the customer just buys from whoever’s cheapest. It’s called carbon leakage, and it’s a real phenomenon, not an invented one. The scheme partially addresses it through Emissions-Intensive Trade-Exposed assistance, which softens the baseline for facilities judged to be genuinely exposed to that competitive risk.
I reckon the EITE carve-out has become more generous than the underlying risk justifies for some sectors, cement being the one I’d point to first. Cement is heavy, it’s expensive to ship long distances relative to its value, and the actual import competition from unregulated jurisdictions is thinner than the lobbying suggests for most of the east-coast market. Alumina and LNG are genuinely trade-exposed in a way that matters. Domestic cement, less so. Treating them the same under EITE assistance is the kind of blanket policy design that’s administratively convenient and economically sloppy.
Where the credits actually flow #
The facilities consistently beating their baselines by the widest margins tend to be the ones that made an early capital bet on electrification or efficiency before the scheme tightened, which means they’re now sitting on a credit surplus they can sell into a market of laggards. That’s the mechanism working as intended, in theory. In practice, the volumes are small enough that a handful of facilities effectively set the clearing price for everyone else’s compliance cost, which is a thin-market problem familiar to anyone who’s watched how the Australian Carbon Credit Unit market itself behaves when a small number of project types dominate supply.
It’s also worth noting the Safeguard Mechanism doesn’t operate in isolation from the broader decarbonisation investment picture. The CEFC and ARENA both fund projects that touch Safeguard-covered facilities: process electrification, heat pump retrofits for industrial heat, early-stage hydrogen trials, and the honest assessment is that without that concessional capital, a lot of the compliance pathway for mid-sized industrial emitters simply doesn’t stack up on a standalone payback basis. Green hydrogen gets raised constantly in this context as the long-term answer for high-heat industrial processes, and it’s worth being precise about where that actually makes sense right now rather than where the spruiking says it does, a distinction covered in detail looking at where green hydrogen genuinely makes sense.
What the next review needs to settle #
The scheme is legislated to continue its decline out to 2030, with a further tightening review built into the architecture. The question that review needs to answer honestly is whether the credit market has enough genuine supply to function as a trading scheme, or whether it’s quietly becoming a disguised carbon tax paid mostly through Australian Carbon Credit Unit purchases rather than process change. Those are very different policy outcomes dressed in the same language. A trading scheme redistributes abatement to where it’s cheapest. A disguised tax just raises the cost of doing business without necessarily buying any additional tonnes of real abatement, depending on how much weight you put on the integrity of the underlying credits: a debate the Clean Energy Regulator and independent reviewers have had running for years now.
My read, for what it’s worth, is that the mechanism is still net-useful, but it’s drifting toward the second outcome faster than its architects intended, and nobody in Canberra has much appetite to say so out loud before an election cycle. The fix isn’t complicated in principle: tighten EITE eligibility criteria sector by sector instead of blanket, and put real teeth into credit integrity auditing so the trading side of the scheme does what it’s meant to. Neither is politically free, which is exactly why neither has happened yet.
The industrial side the grid conversation forgets #
Most of the commentary on Australia’s emissions trajectory still runs through the electricity sector almost by reflex, because that’s where the generation mix changes are visible and photogenic. The Safeguard Mechanism is the less visible half of the ledger, covering roughly a third of national emissions through facilities that don’t get a ribbon-cutting ceremony when they install a new heat exchanger. It deserves more scrutiny than it gets, if only because the compliance cost eventually turns up in the price of cement, steel and LNG contracts, which turns up everywhere else in the economy whether anyone’s watching the baseline curve or not.
Keep an eye on the Clean Energy Regulator’s next facility-level data release. If the credit surplus keeps shrinking the way the last two years suggest, this stops being a compliance story and starts being a competitiveness one, and that’s a different argument entirely from the one Canberra’s currently having.
– Marcus Wren, Editor
Photo by Jaël Vallée on Unsplash