Open in interactive viewer → charts, metric popovers & call review

Golar Is Breaking Its Own Rule to Buy Time — and Has Stopped Talking About the Sale

A 56% sequential EBITDA drop disguised as a build-out: $2.45B of FID'd steel without a charter, $1.6B of fresh liquidity, and a strategic review that vanished from the script
GLNG · Earnings Call · 2026-11-18

A quarter that reads like a trough, priced like a build-out

Golar's Q3 2026 print looks broken at first glance. Operating revenue fell to $84 million from $130 million in Q2, and EBITDA halved and then some, to $56 million from $127 million. The culprit is mechanical, not operational: Hilli finished its eight-year Cameroon charter in late July and went into dry dock. As CEO Karl Fredrik Staubo put it, “Hilli completed her 8-year contract in late July and left Cameroon and is now sitting in a dry dock at Seatrium Shipyard in Singapore in preparation for her 20-year contract in Argentina with startup mid next year.” — Karl Staubo, CEO · 2026-11-18 Net income of $42 million brought the year to $200 million, more than double last year — the residual warm glow of a contract that no longer exists.

So the reported P&L is now almost irrelevant to the equity. What matters is that Golar owns four FLNGs, only one of which (Gimi) is earning today, and it is spending the next three years converting a $4.1 billion market cap into a $17 billion backlog. Contracted run-rate EBITDA is guided to roughly $800 million by 2028 — $150 million from Gimi on a proportional basis, $285 million from Hilli in Argentina, $400 million from Esperanza — rising above $1.2 billion if Unit #4 is chartered on Esperanza-like terms. Against $56 million of quarterly EBITDA, that is the entire bet, compressed into a single sentence about 2028.

The intellectual case management keeps making is that floating beats fixed. land based was the single highest-momentum theme in this call, and Staubo sharpened the boundary with a number rather than a slogan: on capex per tonne and opex per MMBtu, FLNG is competitive up to about 6 to 10 MTPA of liquefaction, and beyond 15 MTPA the physics and the acreage push you back onshore. The proof point he offered is that YPF, Eni and XRG are contemplating a 12 MTPA project in Argentina — floating.

The discipline they promised to keep

Here is where this quarter's story actually lives, and it is a story about a broken promise that nobody on the call asked about. In February 2026, with a strategic review underway and the Chairman on the line, management was explicit that the era of speculative ordering was over.

Instead of going on speculation, number one, because of the different requirements from the various commercial discussions; and number two, for the cash flow profile reasons mentioned by Tor, we've decided to not go on speculation as speculatively as we have previously done and then continue to mature the commercial pipeline before we commit significant capital.

Karl Staubo, CEO · 2026-02-25
Nine months later Golar has taken FID on a $2.45 billion fourth FLNG with no charter signed, acquired the Grace Dahlia as donor vessel from NYK, and — in the same quarter — reserved long lead items for a fifth unit and signed an LOI with Seatrium for another yard slot. The donor vessel strategy itself has tightened: Moss design only, because Moss hulls are poor LNG carriers and therefore cheap conversions, and nothing smaller than Esperanza's 145,000 cubic metres, out to 177,000.

To be fair to them, the sequencing rule they have never broken is FID, not optioning: the fifth boat will not be sanctioned before a contract on the fourth. And the balance sheet has been pre-loaded to absorb the defiance. In one quarter Golar added roughly $1.6 billion of liquidity — a $600 million revolving credit facility, a $500 million unsecured bond, and an expansion of the Hilli leaseback facility to $960 million, with phased draws taking it to $1.2 billion by 2028 to minimise negative carry. Total available liquidity is now about $2.3 billion, before the $1.8 billion of asset level financing being negotiated on Esperanza, which could release a further $1.2 billion. Golar can fund everything it has promised through 2030 without selling equity. That is the real answer to the February objection.

The moat is a calendar date

The competitive claim is not cost, it is timing: earliest available liquefaction capacity globally, end-2029 for Unit #4, a two-to-three-year head start on alternatives. That lead is bought with long-lead equipment, and long-lead equipment is now a contested resource — not against other LNG developers, but against the compute build-out. “We see very strong pressure on long lead items. We are essentially competing for, in particular, compressors and dual fuel engines with the AI data center buildout.” — Karl Staubo, CEO · 2026-11-18 This is the one place where the micro story touches a global one. The market's demand for compute complex has cooled over the last thirty days, with AI-cloud and data-centre capacity names selling off, and long lead materials has de-rated alongside it. Every basis point of AI capex disappointment is, perversely, a tailwind for Golar's input costs and delivery schedule — while every re-acceleration squeezes them further. Golar's moat and its cost inflation are the same variable, pulling in opposite directions.

What stopped being said

Two silences are louder than anything in the prepared remarks.

First, the strategic review — announced in February, probed on the Q2 call with a flat “we will not give any comments on the review itself, neither the outcome nor the timing until we have material information to share” — Karl Staubo, CEO · 2026-08-13 — does not appear in the Q3 prepared remarks at all, and no analyst raised it. Either it is dead, or it is live and the room has stopped asking. Given management's stated view that public pricing understates the backlog, silence here is not neutral.

Second, the buyback, which is where that valuation gap gets expressed. Asked directly, Staubo did not hide: “we do observe that the stock is down around 25% from the peak this year. We don't particularly see a reason as to why the buyback program isn't available.” — Karl Staubo, CEO · 2026-11-18 The buyback program is the only mechanism by which a $17 billion backlog with zero near-term EBITDA converts into per-share value, and it is the closest thing to a catalyst on this call.

Finally, a number for Gimi

The quiet upgrade is operational. For four quarters Golar refused to quantify Gimi's overproduction above contract — in February: “we think more than 3% is fair to assume across the year. Exactly percentages we don't want to commit to right now” — Karl Staubo, CEO · 2026-02-25. This quarter the number arrived: overproduction of 10% to 15% is sustainable, with a further 1% to 3% available from debottlenecking at essentially no capital cost, before any investment-led expansion shared with the upstream partners. Year to date, invoicing sits 12% above the contracted base. Seasonal ambient temperature and maintenance will keep summer lumpy — Q3 managed only 3% — but Q4 is guided higher. That is a small, durable, uncosted uplift on an asset the market treats as a fixed annuity.

Strip out the accounting and the question is simple: is Golar a disciplined allocator of capital entering the steep part of an LNG build-out, or a company that abandoned its own restraint the moment the order book looked tight? The $2.3 billion of liquidity says it can be the former. The missing strategic review, and a CEO calling his own stock attractive while it sits 25% off the high, suggests the market has not yet decided. Gimi's overproduction and the state of the AI capex cycle are the two things that will settle it before Unit #4 has a customer.