Finding: OFAC’s ninth consecutive extension of the deadline to sell Lukoil International GmbH is evidence that the forced-sale mechanism is not producing a sale — not proof that it has failed, but a live test the mechanism is currently losing. Confidence: high that the extension pattern itself is real and documented; moderate that the cause is buyer-side reluctance rather than routine deal friction. Action: read this as a structural finding under observation, not a verdict. The license expires again on 19 September 2026; that date is the next data point, not a deadline this piece is betting against.
1. The nine-extension pattern
On 22 October 2025, the Treasury Department’s Office of Foreign Assets Control designated Rosneft and Lukoil under Executive Order 14024, freezing both companies’ access to the US financial system and, by extension, to the global dollar-clearing infrastructure that underwrites most oil trade. This is established: it is a Tier 1 action, documented in OFAC’s own designation records and corroborated across trade and legal press.
What followed was not a clean wind-down. It was a sequence of temporary permissions to keep negotiating, each one renewed just before the last one lapsed, for exactly the same purpose each time: allow Lukoil to sell Lukoil International GmbH (LIG), the Austrian holding company through which the Russian parent controls most of its non-Russian assets — European refineries in Bulgaria and Romania, roughly 2,000 retail stations across Europe and the United States, a 75 percent stake in Iraq’s West Qurna 2 oilfield, and upstream positions in Mexico, Uzbekistan, and Kazakhstan (stakes in the Tengiz and Karachaganak fields and the Caspian Pipeline Consortium) — to a buyer OFAC is willing to license. Kazakhstan is the one jurisdiction in that list that ends up outside the actual sale, for reasons the Carlyle section below makes explicit.
The sequence, reconstructed from OFAC’s general-license record and confirmed independently through contemporaneous trade-press coverage (Tier 2: FesenkoLaw, Baker McKenzie’s Global Sanctions and Export Controls Blog, the Washington Trade & Tariff Letter), runs as follows. This is established fact; the dates are not in dispute.
| License | Issued | Authorization runs through |
|---|---|---|
| GL 131 (original) | 14 Nov 2025 | 13 Dec 2025 |
| GL 131A | 10 Dec 2025 | 17 Jan 2026 |
| GL 131B | 14 Jan 2026 | 28 Feb 2026 |
| GL 131C | 26 Feb 2026 | 1 Apr 2026 |
| GL 131D | 30 Mar 2026 | 1 May 2026 |
| GL 131E | 29 Apr 2026 | 30 May 2026 |
| GL 131F | 28 May 2026 | 27 Jun 2026 |
| GL 131G | 25 Jun 2026 | 25 Jul 2026 |
| GL 131H | 24 Jul 2026 | 22 Aug 2026 |
| GL 131I | 20 Aug 2026 | 19 Sep 2026 |
Nine renewals — the lettered sequence 131A through 131I — sit on top of the original November license. Each new license issued one to three days before its predecessor expired; there has been no gap in authorization, and no period in the last nine months in which negotiating a Lukoil sale was flatly prohibited. The window’s length drifted before settling: it widened from 38 days (131A) to 45 (131B), then contracted steadily to the 29-to-31-day intervals that have defined every extension since May.
This pattern is worth naming as evidence rather than routine, because “OFAC extended a license again” is the kind of sentence that reads as procedural noise unless the count is made visible. Nine times is not renewal. It is a mechanism that has not once, in ten months, reached the outcome it was built to produce: a completed, licensed sale.
It’s worth being precise about what GL 131I actually authorizes, because the license text is narrower than the headline. It permits negotiation, due diligence, and entry into contingent contracts for the sale of LIG or its majority-owned subsidiaries. It does not authorize the sale itself — any contract negotiated under it must be expressly conditioned on a separate OFAC authorization before money or ownership can actually change hands. The license buys time to negotiate. It does not buy a transaction. That distinction is the mechanical core of why nine extensions have produced zero closings: the license was never designed to force a deal, only to permit one, and permission without a willing, OFAC-acceptable counterparty produces exactly this: an open-ended negotiating window that renews because closing it would either force an unlicensed fire sale or a return to blocked-asset status with no transaction pending.
2. The buyer problem
Two buyers have surfaced publicly since the November 2025 designation, and their trajectories are the clearest available evidence on why nine deadlines have passed without a sale.
The first was Gunvor Group, the Swiss commodities trader, which had a roughly $22 billion deal in place — reported by the Financial Times and confirmed by Bloomberg — covering the full LIG portfolio, contingent on OFAC approval. On 6 November 2025, Treasury killed it before it reached the license stage: a Treasury spokesperson called Gunvor “the Kremlin’s puppet” and stated flatly it would “never get a license to operate and profit” as long as the war in Ukraine continued. (Financial Times, 6 Nov 2025.) Gunvor called the characterization “fundamentally misinformed and false,” citing more than a decade of divestment from Russian-linked trading. (Bloomberg, 6 Nov 2025.) Whichever account of Gunvor’s independence is closer to the truth — and serious reporting on the firm’s ownership history is genuinely mixed — the operative fact for this piece is procedural: OFAC did not evaluate the deal on price or feasibility. It vetoed the counterparty on political grounds before a license request was formally adjudicated.
The second is Carlyle Group, the US private equity firm, which announced an agreement with Lukoil in late January 2026 to acquire most of the LIG portfolio — again excluding the Kazakhstan assets. Lukoil’s own announcement of the Carlyle agreement states that the transaction “does not include the assets in Kazakhstan,” which “will remain to be owned by LUKOIL Group and continue their operations under [their] respective license” — that is, retained directly rather than sold through LIG. Kazakhstan’s government and state operator KazMunayGas are pursuing a separate track, invoking a preemptive right under Kazakh law to buy out Lukoil’s stakes there if they are ever sold; as of this writing Kazakhstan’s Energy Ministry has said it is not currently exercising that right. This is the deal the nine general licenses have been keeping alive. As of the most recent trade-press reporting tied to GL 131H (24 July 2026), no closing had been publicly announced. Reported complications include the scope of the deal (upstream assets like West Qurna 2 sit in a different regulatory category than European retail stations), and reports that Carlyle has explored bringing in Gulf capital — Mubadala, XRG, and International Holding Company have been named in connection with potential minority stakes — to share the financing and, plausibly, the political exposure of the transaction.
We assess with moderate confidence that the extension pattern reflects a genuine mismatch between what sellers can offer and what OFAC will license, not routine M&A friction. The reasoning: nine renewals over ten months is long even by the standard of complex, multi-jurisdiction energy asset sales, and the one deal OFAC’s own rhetoric killed outright (Gunvor) was killed for who the buyer was, not what it offered. That is a data point about the licensing agency’s revealed preferences, not the market’s. We do not know — and no Tier 1 or Tier 2 source we found states — whether Carlyle’s structure has stalled over price, over the political optics of a US firm absorbing sanctioned Russian assets even at a discount, over the complexity of separating upstream from downstream assets, or over something OFAC has not made public. That gap is real and we are naming it rather than filling it with inference.
3. The steel-manned case for the mechanism working
Before extending the No. 17 argument, the enforcement view deserves its full statement, because it is not naive. Secondary sanctions do not need to force an immediate sale to be doing their job. Lukoil’s FY2025 IFRS financial statements report a full write-off of its investment in Lukoil International — an impairment loss of 1.66 trillion rubles — meaning Lukoil itself has already accounted for LIG as a loss, sale or no sale. On this view, nine extensions are not evidence of failure; they are evidence of a company being slowly bled of a $20 billion-plus asset base while its own leadership can find no buyer OFAC will accept, which is precisely the coercive pressure sanctions are meant to apply. A forced sale at a firesale price to a US private equity buyer, on a timeline dictated by Washington, is a win for the sanctions regime even if it takes eighteen renewals instead of one. Slow strangulation is still strangulation. Judged against “has Lukoil been meaningfully deprived of its Western asset base,” rather than “has a closing occurred by a specific date,” the mechanism has a real claim to be working.
4. Connection to the Economic D-Day thesis
Sounding No. 17 argued that Trump’s 19 August threat against Iran’s sanctions-evasion architecture was a bet that secondary sanctions retain the coercive power they had in 2012, when the SWIFT exclusion cut Iranian oil revenue by roughly half — a result that worked specifically because dollar-clearing had no substitute. That piece’s structural question was whether enforcement in 2026 can still reach a trade corridor built, this time, to route around the dollar system entirely: Chinese teapot refineries in Shandong settling Iranian crude through the Bank of Kunlun via China’s Cross-Border Interbank Payment System, using yuan-denominated and barter-like clearing arrangements that never touch a US correspondent bank. Treasury’s own April 2026 warning to banks about “teapot” refinery exposure, and its follow-on sanctions against nineteen tankers and a Dalian refinery under “Operation Economic Fury,” are Tier 1 evidence that Washington knows the corridor exists and is trying to reach it anyway — with, so far, incomplete success: China’s government moved in May 2026 to block enforcement of those sanctions against five of its refiners on its own soil, a direct assertion that its financial system is now large enough to shelter its own actors from a foreign asset freeze.
Lukoil is a different case in an important respect: it runs through the dollar-and-euro system, not around it. LIG’s assets are European refineries, Iraqi oilfields, and US-adjacent retail stations — precisely the kind of asset base that cannot simply reroute through CIPS the way a Shandong refinery’s yuan payment can. That is exactly why Lukoil is the cleaner test of the classic mechanism, stripped of the alternative-payment-rail complication that muddies the Iran case. If secondary sanctions can force a divestment anywhere in 2026, it should be here, where the target has no yuan off-ramp and no CIPS alternative — its entire foreign asset base sits inside the jurisdictions the dollar system actually reaches. Nine extensions without a closed sale is therefore a harder finding than it would be in the Iran case: it suggests that even where the enforcement architecture has no structural workaround to contend with, the deadline-and-license tool is still not compelling a transaction on Washington’s timeline. The friction, on the Lukoil evidence, is not only that targets have found a way around the rails. It is that even with no way around the rails, the mechanism is slow, contested, and not obviously terminal.
Read together, the two cases suggest a mechanism under stress from both directions at once: where an alternative payment architecture exists (Iran, via CIPS), sanctions face a structural bypass; where no bypass exists (Lukoil, via LIG’s Western-jurisdiction assets), sanctions still face a buyer-acceptability bottleneck that nine deadlines have not resolved. Economic D-Day asked whether enforcement could follow the announcement into unmapped territory. The Lukoil case asks the prior question: can it close the deal even on mapped territory. Nine extensions in, the answer is not yet.
Prediction 1 — GL 131J: We assess it likely (moderate confidence) that OFAC issues a tenth extension — a GL 131J — before or shortly after GL 131I expires on 19 September 2026, on the same pattern: issued in the final days of the prior window, covering roughly 30 days, with no public indication of an imminent closing as of this writing. The basis for this assessment is the unbroken nine-for-nine track record of renewal-before-lapse and the absence, in any Tier 1 or Tier 2 source reviewed for this piece, of a stated OFAC deadline that is not itself a license expiry date.
Prediction 2 — Carlyle closing: We assess with low confidence that the Carlyle transaction closes, in some form, within the next two license cycles (by year-end 2026) — low confidence because the public record shows an agreement in principle from January but no resolved structure, and because the Gulf-capital reporting suggests the deal may still be evolving in scope rather than approaching signature.
Resolution: GL 131J issuance (or its absence) by 20 September 2026, and the Carlyle transaction closing (or its absence) by 31 December 2026 — both logged to the publication’s prediction ledger and scored as they resolve.
We do not know whether a further OFAC rejection of a specific buyer or structure — a second “Gunvor moment” — is coming. Nothing in the sourced record points to it, but nothing rules it out either, and the Gunvor precedent shows OFAC is willing to kill a deal on counterparty grounds after most of the negotiating work is done. We flag that possibility here as an open question worth watching, not as a Ledger entry: unlike the two predictions logged above, it has no bounded resolution window and no defined condition that would let a non-occurrence be scored as correct rather than simply unresolved.
If a tenth extension arrives with still no closing, the finding in this piece hardens from “mechanism under live test” toward “mechanism not producing its designed outcome.” If a closing occurs, the piece’s core claim — that repeated punts signal no acceptable buyer, not routine process — will have been wrong, and we will say so plainly when it happens.