This weekend, the US and Iran paused their direct exchanges after two weeks, and you’d think that means things are cooling off. Instead, on a separate front, the Houthis went ahead this weekend and hit Saudi Arabia directly, their first strike on Saudi oil infrastructure since 2022, after declaring a naval blockade against Saudi Arabia last week. Whatever is happening between Washington and Tehran, the Houthis are clearly following their own logic.

Early Saturday, Houthi missiles and drones struck Jizan while attempting to hit Yanbu. The Jizan refinery caught fire. At Yanbu, two ballistic missiles aimed at the oil installations there were intercepted. The Houthis framed it as retaliation for Saudi strikes on Hodeidah and Kamaran Island.

Yanbu is the part that should make everyone pay attention, and it’s already been the center of the story for a couple of weeks now. Once the Houthis declared their blockade against Saudi Arabia and blocked the direct route out through Bab-el-Mandeb, Yanbu became the only place Saudi Arabia could still push its Red Sea crude, roughly 3.5 to 4 million barrels a day, nearly all of it bound for Asia. It is now handling around 92% of Saudi Arabia’s seaborne crude exports right now, basically the country’s only real way to move crude out by sea with Hormuz effectively closed as well. Which is why, before this weekend, most of the conversation around Yanbu was a plumbing problem rather than a supply problem.

And it really is a very interesting plumbing problem, because it explains why there’s a 1970s pipeline suddenly back in the news. SUMED exists because of an earlier version of this same situation.

The Suez Canal was closed for eight years after the 1967 war, and while it was closed the industry built bigger vessels to make the long way around Africa with a lower cost per barrel. By the time the canal reopened in 1975, a big chunk of the world’s tanker fleet had grown into VLCCs too deep to fit through it anyway. So a consortium of Egypt, Saudi Aramco, Kuwait, the UAE and Qatar built a pipeline instead, Ain Sukhna on the Red Sea side to Sidi Kerir on the Mediterranean side, roughly 320 kilometers, which has been operating since 1977 with a capacity of around 2.5 million barrels a day. It already had one moment earlier this year, back in March, when Egypt’s own oil minister publicly offered it up as a release valve once the Strait of Hormuz shut. With Bab-el-Mandeb blocked too, it stopped being a convenient option and became one of the only ways out.

The reason this pipeline is needed is that a VLCC fully loaded cannot transit the Suez Canal, because of its draft. So getting Yanbu crude to Asia the long way around means solving that problem first, and there are basically four ways:

1. One VLCC can shuttle between Yanbu and Ain Sukhna, discharging into the pipeline, while a second VLCC sits at Sidi Kerir on the Mediterranean side loading the cargo before heading off around the Cape. Looks like a great idea, except the Mediterranean isn’t a VLCC trading region day to day, so finding that second ship means either getting lucky or ballasting one in from the Atlantic, which can take weeks, and either way you’re capped by the pipeline’s 2.5 million barrel a day ceiling.

2. Or a single VLCC can do the job, loading at Yanbu, discharging just enough (around 800.000 barrels) at Ain Sukhna to get under the Suez draft limit, transiting the canal lightered, then picking the same crude back up at Sidi Kerir once it’s come through the pipeline, before topping up its cargo again before heading out around the Cape. Same ship the whole way, no second VLCC needed, and the pipeline gets used about as efficiently as it can be.

3. Or a VLCC part loads at Yanbu and transits Suez lightered like above, but instead of using the pipeline, an Aframax carries the rest through the canal on its own, since an Aframax doesn’t have the draft problem, and the two meet up in the Mediterranean to do a ship-to-ship transfer. This one skips the pipeline capacity question entirely, but you’re paying for two ships & Suez Canal tolls plus the STS itself. Very expensive.

4. Or you just charter Suezmaxes directly, since they can transit the canal fully loaded with no lightering and no pipeline involved at all. It is the simplest option operationally if the Suezmaxes do the full voyage to Asia, but more expensive than the first and second option by my numbers. You could also do an STS in the Mediterranean Sea from two Suezmaxes to a VLCC, but this option is also expensive.

I have already seen two vessels on subs for the first option (the first one at $17m lumpsum from Sidi Kerir to Korea) and I guess that we could end up seeing both the first and fourth options being used simultaneously in different proportions depending on the charterer and the vessel availability.

Today, the situation escalated again. Houthi and Iranian state media have begun reporting fires at Abqaiq, the largest crude stabilization and processing facility in the world.

Saudi Arabia’s own statement only confirmed intercepting drones launched from Iraqi territory aimed at oil facilities, it didn’t confirm Abqaiq damage. The images going around show smoke and flaring, and flaring on its own doesn’t tell you a plant took real damage. So I treat the Abqaiq claims as a possibility for now, not as confirmed fact.

Everything about the four options assumed the crude itself, and the terminals loading it, are still working, that the problem is purely getting ships from A to B. That’s a cost problem, ugly but manageable, and worth understanding on its own terms. But the real question raised this is whether the Houthis have actually gained the capability to reach into Saudi Arabia’s oil infrastructure and materially affect its export capacity itself. If the answer is yes, this stops being a shipping story and becomes a supply story, and those are priced completely differently.

From my tweets, most people probably file me under “bear.” And fair enough, there’s some truth to that. My base view is that there’s a lot of oil in the world. Way more than the price action of the last few years would suggest. The only reason prices have held up (in nominal terms, if you adjust for inflation it gets worse) is that OPEC has been pulling a huge chunk of supply off the market on purpose. And I’m sorry, but a market where you artificially withhold a big slice of production cannot be called a market in deficit. That’s not a deficit, that’s a cartel doing its job.

Some numbers on the cuts, because this gets thrown around a lot: OPEC+ has had roughly 1.65 million barrels a day held back since April 2023, a cut that’s been extended out to the end of 2026. On top of that there was another voluntary layer of 2.2 million barrels a day announced in late 2023, led by Saudi Arabia alone taking 1 million b/d off the table. That second layer got unwound gradually through 2025, but even after releasing close to 2.9 million b/d back into the market since April last year, OPEC+ still had around 3.24 million b/d of cuts standing when they paused hikes for Q1 2026. The pause ended, and since then it’s been hike after hike after hike, every single month, April, May, June, July, and now August too. Five in a row. Call it another 1.2 to 1.4 mbd (theoretically) back in the market on top of what was already released before the pause. The number of barrels supposedly “cut” keeps shrinking every month and it only ever moves one direction.

Then there’s the UAE, which is the best proof of the whole argument I could ask for. On May 1st this year they walked out of OPEC after 59 years in it. Why. Because their quota was capped at 3.4 mbd while ADNOC’s real capacity sat something like 4.5 mbd. You don’t quit a 59 year old club over a rounding error. You quit because you can pump more oil than they’re letting you and you’re done pretending you’re happy about it.

And if you zoom out a bit further, the incentives of the Gulf producers point in exactly one direction: pump as much as you can, as fast as you can. These are countries sitting on decades of cheap barrels, watching the world talk more and more about peak demand and energy transition every year (which is something I don´t share). But if you genuinely believe your oil might be worth less in twenty years than it is today, you don’t hold it back out of politeness. You sell it and reinvest elsewhere.

So that’s the bias, out in the open. But besiades that I have never been short oil or oil related names, and In have always stated that I would love a $75-80 range in which everyone could be happy: producers, refiners, customers… but the equilibrium state is unfortunately unstable. Now here’s where I actually learned something.

2026 started and every signal pointed to a big glut coming. That was the consensus, that was somehow my view too. And yet oil equities just kept grinding higher, no real pause, especially once tension with Iran started building. A few sharp people caught this early and said so out loud, PamplonaTrader being the obvious one who flagged it in my feed. I’ll be honest, this is one of the calls I didn’t get fully right. What I did have during that stretch was a big position in Kosmos, which was really more of a company-specific turnaround story than a macro oil call, and a lot of credit there goes to Gabriel Castro for the work he put into that name. I also had a decent Serica position running alongside it, which worked out well too. And Golar. I was very long Golar, much less now.

https://x.com/ayusovalue/status/2026934105028399475?s=46

Then the war actually started, and weirdly enough oil equities went flat for a while. I sold both Serica and Kosmos and played some of the dips with 0dte bull put spreads on WTI futures.

At the same time, separately, I was building a large position in Constellation Oil Services (the subject of the last articule I wrote). Not because of the war directly, but it’s pretty clear the Brazilian offshore sector is a beneficiary of what’s happening in the Middle East.

That’s when things got genuinely interesting. China started cutting its import volumes. Trump started giving off TACO signals again. And then the MOU landed, along with a waiver on Iranian oil, and that sent oil straight down toward 70. Not only the price, the whole futures curve flipped into contango. I’ll keep repeating this because it matters, the shape of the curve tells you almost more than the spot price does.

For a few weeks, if you actually added it all up, inventory releases coming back into the market, Chinese demand craterind, the various bypass routes, actual transit volumes through Hormuz, the picture was genuinely bearish for oil. But then tension picked back up again, the attacks around Hormuz started again, and the US pulled the Iranian oil waiver. That was the bottom, and I said as much at the time on Club del Value Discord (unfortunately only in Spanish). Brent closed at 72.85$ that day.

There was also an interesting event that I also commented on in El Club del Value Discord: two pieces of negative news in principle for the price of oil, yet it held, a strong bottom signal as well. OPEC+ raising quotas and Saudi OSPs reduced to 2020 lows…

Transit numbers through Hormuz started falling off a cliff right around then too. On top of that, the main contrarian indicators I watch lit up hard, and I posted that on Twitter as it happened.

That was the moment to buy leveraged oil names, not after oil had already run back to 100. Leaning on everything I already knew about Kosmos, and helped by a genuinely strong operational update the company put out right around then, I bought shares again.

Along with that, I put on a combo I also shared with Club del Value at the time: sell the $2 put, buy the $2.5 call, sell the $3 call, all in October. Structured so you actually get paid $0.1-0.15 per contract to take on the upside in Kosmos from $2.5 to $3.

In addition, I publicly disclosed buying more Kosmos one week ago at $2.17. It closed yesterday at $2.63, i.e. up 21% from that price.

So, bull or bear. Honestly, wrong question. I still think there’s too much oil in the world long term, and I still think the deficit story is mostly a supply-management story wearing a scarcity costume. But none of that stopped me from buying leveraged oil equities at what turned out to be the low, because the setup in front of me at that moment said buy, not because some macro thesis said so. The lesson isn’t that I flipped, the lesson is that flipping when the data tells you to isn’t a betrayal of your own view, it’s the whole job. Marry a framework, not a position. Everyone gets things wrong from time to time; I make a lot of mistakes and bad calls too. The people who get hurt in investing are the ones who refuse to let go of the need to be right.

Disclaimer: The views expressed in this article are solely those of the author and are provided for informational and educational purposes only. Nothing contained herein should be construed as investment, financial, legal, or tax advice, nor as a recommendation or solicitation to buy, sell, or hold any security or other financial instrument.

The author may hold positions in some of the securities discussed. Readers should conduct their own due diligence and, where appropriate, consult a qualified financial adviser before making any investment decisions. Past performance is not indicative of future results.

This isn't meant to be a detailed investment thesis; that will come later if I have the time. But since the company has just announced the refinancing of its bonds, and it's my largest position, I wanted to share a few thoughts.

The Company

Constellation Oil Services Holding is one of the main offshore drilling operators in Brazil, running a fleet of semisubmersibles and drillships mostly under Petrobras contracts, plus meaningful private-client work (Shell, Total, Brava Energia). It has one of the highest revenue visibility in the offshore drilling space: around 87% of 2026-28 revenue estimates already sit on firm backlog.

It holds roughly 25% market share in Brazil and ranks first in Petrobras’ own performance table (”Sondópolis”), with 99% operational uptime in Q1 2026.

The July 2026 refinancing: more than a coupon repricing

Yesterday, Constellation priced a USD 650 million senior secured bond due 2033, fixed coupon 7.7% a year, paid semi-annually. Proceeds plus cash on hand will pay off the existing USD 650 million 9.375% senior secured notes due 2029 in full. The coupon drop, 167.5 basis points, saves around USD 10.9 million a year in cash interest at the same face value. Not a massive number in absolute terms, but a clear sign the company’s credit profile and access to capital markets have improved a lot since that last bond was priced. The book was heavily oversubscribed, highlighting strong demand from investors to provide financing to the company, and a 320bps spread is very decent.

What’s actually interesting here is:

• 7-year maturity, chosen on purpose. Given the company’s contract coverage and how backlog rolls off, an amortizing structure makes more sense than a longer bullet against assets that are depreciating. Management frames it as an optionality window, a chance to reprice or refinance starting year 3, or just hold the bond to year 7.

• The net leverage covenant is gone. It existed on the old notes; it doesn’t exist anymore. A real change to the credit profile, even though the company keeps its own internal leverage discipline (more on that below).

• The collateral changed too. Before, the security package was rig contracts (not the rigs themselves) plus the company’s outstanding receivables. Now it’s the real assets, the rigs Constellation owns outright, 6 of them pledged, down from 7 before. (Interesting for bondholders).

• Extra debt basket. The company keeps a general basket of USD 100 million sized for a potential RCF plus a main basket of USD 200 million, about USD 300 million of incremental debt capacity available under the same collateral, without reopening the whole security package. That’s dry powder for growth or M&A.

• Old 2029 bond amortization, for context: 10% callable at 103, the rest via make-whole premium at T+8, standard high-yield stuff that explains why refinancing now made sense. Average price paid was 105.6.

Capital allocation, the real point of this article

This is the core of the whole thesis. Management said it clearly: revenue visibility (that 87%+ backlog coverage) combined with net debt under 1.25x gives them confidence to step up the dividend in a sustainable way going forward.

Right now the ordinary dividend is USD 100 million a year (around 8% yield), started at the beginning of 2026, the first year the company formalized this policy. The stated priority order for capital is: generate cash, service debt, spend capex to keep the rigs in good shape, then pay shareholders. On net debt, management itself said they’ll run “closer to one” (roughly 1-1.25x NIBD/EBITDA), and they see that as sustainable.

Put simply, the company is already generating the cash, the debt is where management wants it, the new bond has lighter covenants and the only thing missing is the formal announcement of the bigger dividend.

I think it is quite reasonable to assume the company will be able to increase the ordinary dividend to USD 200 million in 2027. With leverage expected to remain comfortably within management’s target range, strong earnings visibility backed by long-term contracts, and limited capital requirements beyond maintenance capex, doubling the current payout would still leave the balance sheet conservatively positioned while offering shareholders a yield that stands out within the sector.

Fleet and backlog

Constellation just added over USD 1 billion of new backlog by negotiating three contracts with Petrobras. (Arctic Charts)

But the two names that matter most for the near to medium term are Lone Star and Atlantic Star.

Lone Star. Petrobras pulled basically every rig off the open Brazilian market for 2027, which makes Lone Star the only deepwater option left for the private (non-Petrobras) market next year. That goes back to a strategic call made years ago to dedicate one rig to the private market, a decision that helped Constellation build its reputation with Shell, Total, and Brava. Lone Star is now on its fourth project for Brava Energia and is likely to get extended precisely because there’s no other rig available for 2027, and the company is also negotiating two more projects outside of Petrobras for that same unit. Hard to think of a better pricing setup for a single asset.

Atlantic Star. This rig (a semisub built in 1976) is going through a career shift. It’s now fully configured in moored mode for shallow water work, moving away from long-term Petrobras contracts to go after a fast-growing market driven by independent operators that need specialized units for workovers and P&A (well abandonment) in shallow water. What matters for investors here is that management has deliberately simplified how it wants this rig read, calling it “pure upside.” The 2026 EBITDA guidance (USD 350-385 million) assumes zero contribution from Atlantic Star. So any new contract this rig lands is straight upside on top of guidance already given, not something already baked in.

Barriers to entry in Brazil

Anyone wanting to bring a foreign rig into Brazil needs to spend USD 100-150 million (plus about 6 months of idle-time opportunity cost) on what the industry calls a “tropicalization package,” on top of standard mobilization costs.

On top of that, there’s a bigger structural point that applies to the whole sector, not just Brazil. Management doesn’t expect any new deepwater drilling rigs to be built from scratch and enter the global market. Neither do I. Building one today costs over USD 1 billion and takes 3 to 4 years, and would need a long contract (6 years) at rates above current market levels (around USD 700,000/day) to justify the spend, basically a 10-year total commitment (build plus contract) that a lot of owners and operators just aren’t willing to make anymore. That keeps the existing installed fleet scarce and valuable across the whole sector, Constellation included.

The Brazilian market beyond Petrobras

This is the part I find most interesting for the medium term, because it doesn’t depend only on Petrobras.

Brazil’s oil regulator (ANP) auctioned rights to 34 offshore exploration blocks to 9 different energy companies last year (2025). Companies like Chevron and ExxonMobil paid big amounts in upfront signing bonuses to drill in these areas. There’s a big focus on the Equatorial Margin (Brazil’s north/northeast offshore coast), a new exploration frontier for the country. To put it in perspective, Brazil’s total offshore reserves sit at around 46 billion barrels of oil today. A recent report from Brazil’s Energy Research Office (EPE) says the Equatorial Margin alone could hold up to 41 billion barrels. That would nearly double the country’s offshore reserves. We’re only at the stage of drilling the first exploratory wells, but it’s encouraging to see the Brazilian offshore oil and gas market keep maturing. The return of majors like ExxonMobil and Chevron to Brazilian deepwater projects, after a period of relative absence, backs up that positive read.

Shallow water operations are also benefiting from Petrobras’ divestments over the past few years. Independent operators like Trident Energy, Karoon Energy, and Brava Energia are now thriving in that segment, and that’s exactly the market Atlantic Star is repositioning toward (workovers and P&A for these independents). It explains why management sees this rig as a genuine growth opportunity rather than just a tail asset in the fleet.

All of this sits on top of Petrobras’ 2026-2030 five-year plan (released November 2025): USD 109 billion of total portfolio investment, with E&P breakeven now down to around USD 25/bbl Brent (from around USD 28/bbl before). 15 FPSOs already awarded, 17 more in the pipeline (Arctic chart), and 40 new exploration wells planned for 2026-2030. Brazilian rig demand has grown from a trough of 12 floaters in October 2019 to 33-35 today, stable over the last 18 months.

The offshore market: supply, demand, and oil price

Everything above only works if the offshore drilling market itself stays tight, so it’s worth stepping back from Constellation for a minute and looking at the bigger picture.

On supply, there’s barely anything new coming. Across floaters, jackups, harsh environment semis and tender rigs, the order books are close to empty. Tier 1 and Tier 2 drillships are already running around 92% utilization, Tier 1 specifically at 98%, and forecasts put utilization at 92% in 2027 and 96% in 2028, above the last cycle’s peak. Dayrates for Tier 1 drillships are pointed at USD 400-450k a day in 2027 and USD 450-500k in 2028. Jackups are recovering more slowly, Saudi Aramco is gradually bringing back suspended rigs and Mexico’s tendering is picking up as Pemex’s payment issues ease, with total jackup utilization forecast to reach 89% by 2028. Harsh environment semis are the tightest pocket of all, the fleet is down 49% since 2015 and 2026 demand coverage already sits around 95%, dipping to 67% for 2027 before options push it back toward 99%. None of this turns around quickly. Building a new deepwater rig costs over USD 1 billion and takes 3 to 4 years, and needs a 6-year contract at rates above today’s market to make sense, basically a decade-long commitment most owners aren’t willing to make anymore, which is exactly what Constellation’s own management said in the meeting behind this piece.

On demand, Brazil alone is a big part of the story. Petrobras’ 2026-2030 five-year plan puts USD 109 billion into the portfolio, with 15 FPSOs already awarded and 17 more in the pipeline, plus 40 new exploration wells planned. Add the ANP’s 34-block auction and the Equatorial Margin exploration wave described above, and Brazilian demand alone should keep growing rather than just holding steady. Globally, Rystad expects 60 new projects to get sanctioned in 2026, with greenfield spending reaching 55% of total offshore E&P spend, and majors like Chevron have flagged a 50% jump in planned exploration spend. Norway’s big three, Equinor, Aker BP and Var Energi, remain committed to heavy investment through the decade too.

Now the part that ties it all together, does the oil price actually support all this spending? According to Rystad Energy, the breakeven for offshore deepwater projects sits around USD 43 a barrel of Brent as of 2026, one of the cheapest sources of new oil supply out there, even as the average breakeven across non-OPEC projects has climbed to USD 47 a barrel. Petrobras’ own breakeven is even lower, around USD 25 a barrel, which is exactly why its five-year plan keeps growing instead of shrinking. Rystad’s own long-run estimate for the oil price needed to balance 105 million barrels a day of demand by 2030 is about USD 55 a barrel, still comfortably above the deepwater breakeven.

And here’s where the futures curve actually matters for this thesis. As of 23 July 2026, front-month Brent is trading close to USD 100 a barrel, pushed up by the Iran-Israel-US conflict and the disruption around the Strait of Hormuz. That’s spot price noise from a war, and it won’t last forever. But the curve itself tells you what the market expects once that noise clears, it’s in steep backwardation, dropping to around USD 77 a barrel by December 2027 and settling into a long, flat structure around USD 67-70 a barrel for contracts running out into the mid-2030s as the following picture shared by Rory Johnston (@Rory_Johnston) shows, btw one of the best oil analysts out there in my opinion.

Even that settled, post-crisis level sits USD 24-27 a barrel above the deepwater breakeven, and close to three times Petrobras’ own breakeven. The current war premium doesn’t need to hold for offshore investment to keep making sense. Even if Brent fully unwinds back down to its long-run curve level, offshore projects in general, and Petrobras’ projects in particular, clear their cost of production with a wide margin. That’s the real reason Constellation’s backlog and dividend visibility don’t depend on oil staying at USD 100, they depend on it staying above roughly USD 45-50, which the market itself is already pricing in for years to come.
Note that the futures curve is a poor predictor of future spot prices, but it remains an important reference for sanctioning long-term projects.

M&A and consolidation

Management would consider M&A, especially if it’s cash accretive, with a clear priority to pay in equity so it doesn’t disturb the capital allocation guide above. In other words, the dividend policy comes first, M&A can’t put that at risk. Constellation sees consolidation in the Brazilian sector as positive for the industry whether or not it takes part itself, and is open to talking to potential counterparties. There’s no stated restriction between deepwater and shallow water, even though today they only have one rig dedicated to shallow water, since most Brazilian demand is still deepwater.

With 25% market share and the top spot in the performance ranking, management argues other operators would benefit from Constellation’s ability to add rigs without heavy infrastructure investment. Last year they already ran a “light asset strategy,” adding two rigs to the portfolio without a proportional jump in leverage, which improved scale and cut costs. In their own words, it’s “an alternative to keep growing” without stretching the balance sheet. Worth keeping an eye on Ventura Offshore here, the most obvious consolidation target given how established Constellation already is locally.

Valuation

At 141 NOK, the valuation looks increasingly difficult to justify. The adjusted EV/EBITDA multiple compresses from 4.1x in 2026e to 3.2x in 2027e and 2.6x in 2028e, driven solely by EBITDA growth and lower leverage. Even applying a conservative 4.5x multiple to 2027e EBITDA would imply a share price of roughly 200 NOK, despite the stock still trading well below the 6–8x EV/EBITDA range of peers.(Arctic Chart)

Risks

The main ones here: concentration in a single country and, to a large extent, a single client (Petrobras), though diversification toward private clients via Lone Star and toward shallow water via Atlantic Star helps offset that a bit. A fleet that doesn’t carry the newest technology, though as explained that’s offset by the Petrobras-funded upgrade model. Execution risk in repositioning Atlantic Star toward a still-young workover/P&A shallow water market. And the commodity cycle itself, this whole story depends on Petrobras and the majors keeping up the level of offshore investment that’s currently taken for granted.

Conclusion

Constellation is going into this refinancing with the balance sheet where it wants it, close to 90% of the next three years’ revenue already contracted, two assets (Lone Star and Atlantic Star) offering upside that isn’t even in the guidance, and a Brazilian market, inside and outside Petrobras, that keeps expanding rather than winding down.

Management already put the key sentence on the table themselves: revenue visibility and debt under 1.25x give them confidence to raise the dividend sustainably.

Disclaimer: this piece reflects my personal view as an investor on Constellation Oil Services and isn’t investment advice. I hold a position in the company. Do your own due dilligence.

Know the refinery, the actual physical plant that turns crude into the products people burn, drive on, and fly with. His point was obvious in hindsight: crude oil has no value on its own. Nobody wants a barrel of crude. What people want is gasoline, diesel, jet fuel, and a dozen other refined products. Crude is just the raw material standing between a producer and those products, and the machine that does the converting, the refinery, is where the real economics of oil emerge. If you don’t understand that machine, you’re trading a derivative of something you’ve never actually looked at.

So this is the first post in a short series on refining, not a technical deep dive for engineers, but the version a trader, an analyst, or an investor actually needs.

Today: the basics.

Crude is not one thing

The first mistake newcomers make is treating “crude oil” as a single fungible commodity, the way you might think of gold or wheat. It isn’t. Crude varies enormously by grade; some is light and easy to process, some is heavy and sulfurous. Two numbers do most of the work in describing a barrel: API gravity (roughly, how light or heavy it is: higher API means lighter) and sulfur content (sweet crude is low-sulfur, sour crude is high-sulfur and more expensive to clean up).

Why does this matter? Because no refinery can run every crude equally well. Refineries are engineered, often over decades and billions of dollars, around a specific diet of crude qualities. A refinery built for light sweet crude will choke, economically, if not literally, on a heavy sour barrel it wasn’t designed for, and vice versa. This is the first thing “know the refinery” means in practice: before you can say what a barrel of crude is worth to a specific buyer, you have to know what that buyer’s plant is actually built to process.

The first cut: distillation

Every refinery starts with the same basic step: the crude distillation unit, or CDU. Crude oil is a mixture of thousands of different hydrocarbon molecules, and the CDU just sorts them. Crude is heated and fed into a tall column, and because different hydrocarbons boil at different temperatures, they separate into layers as they rise: the lightest gases and naphtha come off near the top, kerosene and diesel-range material in the middle, and the heaviest, thickest material, atmospheric residue, stays at the bottom.

That residue often goes through a second round in a vacuum distillation unit, which uses reduced pressure to pull more value out of it without the extreme heat that would otherwise crack the molecules apart uncontrolled. Together, atmospheric and vacuum distillation are just sorting machines. They don’t upgrade anything. Whatever value is locked in that heavy bottom-of-the-barrel material is still there, the question the rest of the refinery has to answer is what to do with it.

Complexity: the question that actually sets margin

This is where refineries stop looking alike. A simple “topping” refinery does distillation and not much else. It sells whatever comes off the column, including a lot of low-value heavy residue that’s barely more useful than the crude it started with. A complex refinery, by contrast, has a second layer of units: a fluid catalytic cracker (FCC), a hydrocracker, a coker, whose entire job is to take that heavy, low-value residue and break the long hydrocarbon chains into shorter, more valuable ones: turning gasoil into gasoline, turning residue into diesel, turning tar-like bottoms into something sellable at all.

The industry has a shorthand for how much of this upgrading muscle a refinery has: the Nelson Complexity Index. It’s a single number, but it maps directly to something a trader cares about enormously: how much a refinery can afford to pay for a cheap, heavy, sour barrel versus how dependent it is on expensive light sweet crude. A high-complexity refinery on the US Gulf Coast can run discounted heavy Canadian or Latin American crude and still turn out a product slate dominated by high-value gasoline and diesel. A simple refinery has no such option, its economics are tied much more tightly to the price of light sweet crude, because it has no way to fix what a heavy barrel gives it.

This is the second thing “know the refinery” means: two refineries looking at the same crude offer can value it completely differently, and the gap between them is complexity.

Why the crude price alone tells you almost nothing

Here’s the part that changes how you read the market. Financial media reports the price of Brent or WTI as if it were the whole story. It isn’t. What a refiner actually cares about, what determines whether running that refinery is profitable this week, is the spread between the cost of the crude going in and the value of the products coming out. The industry calls this the crack spread, and the most commonly quoted version, the 3:2:1, approximates it as: three barrels of crude in, yielding roughly two barrels of gasoline and one barrel of diesel out, priced at market.

The chart below from Alexander Stahel compares front-month WTI crude prices with the gasoline crack spread over the past year, highlighting their close relationship and recent divergence.

When the crack spread is wide, refiners want to run flat out, crude demand is strong regardless of what the outright crude price is doing, and refiners bid crude prices up chasing that margin. When the crack spread collapses, because product demand is weak, or too much refining capacity is competing for it, refiners cut runs, and crude demand softens even if the crude price itself hasn’t moved. A trader watching only the crude tape misses this entirely. A trader who understands the refinery is watching the spread, not the level, and that’s usually where the real signal is.

Products aren’t interchangeable either

One more layer: what comes out of the refinery isn’t just “gasoline” or “diesel” in the abstract, it has to meet specifications. Octane rating for gasoline, cetane for diesel, sulfur limits that have gotten steadily tighter (the 2020 IMO rule capping marine fuel sulfur at 0.5% reshaped an enormous slice of the global product market overnight). Meeting spec requires blending, i.e., combining different refinery streams in the right proportions, and getting that blend wrong, or being unable to make it at all, is its own source of value and constraint. A refiner that can only make high-sulfur fuel oil in a world that increasingly won’t buy it has a very different economics than one that can blend its way to compliant product.

Back to the lunch table

So what did that trading director actually mean? I think it was this: crude trading looks, from the outside, like a bet on a single price: will Brent go up or down. But underneath that price sits a chain of machinery, and every link in that chain: crude quality, distillation, complexity, crack spreads, product specs, is where value is actually created, destroyed, or arbitraged. A trader who only watches the crude number is playing checkers. A trader who understands what the refinery does with that barrel, and why, is playing something closer to chess.

This is the thing many people keep missing: oil isn’t a scarce commodity in the way gold is. It’s abundant, fungible, and sitting in the ground or in tanks in a dozen countries. What’s scarce, what’s genuinely hard to substitute on short notice, is the ability to move it; a tanker, in the right place, insured, flagged, and willing to sail through a war zone. Barrels are a commodity. Freight is the constraint. And the constraint is what sets the price at the margin.

The market created by logistics

This isn’t a new phenomenon; it’s close to the founding condition of how oil actually trades. The modern spot market for crude was built by vessels being in the wrong place.

When the Suez Canal shut in 1967, tankers bound from the Gulf to Europe suddenly had to sail around the Cape of Good Hope, tripling the voyage and blowing up freight economics overnight. Cargoes that had been priced under long-term contracts with an assumed freight cost became commercially nonsensical the moment the shipping math changed underneath them. Companies started swapping and reselling cargoes to optimize who was closest to what, and a market in physical barrels priced independently, cargo by cargo, started emerging almost as a side effect of a shipping problem. Rotterdam grew into a spot hub because it was where flexible, incremental cargoes naturally landed when the freight math didn’t fit the old contract structure.

The North Sea told the same story a decade later. Brent crude, loaded from a single offshore platform onto tankers on a rolling schedule, wasn’t well suited to the old system of annual term contracts either: cargo dates slipped, ships were delayed, positions needed to be traded before anyone knew the exact loading date. So traders started buying and selling the right to a cargo (15-day Brent) before it even existed as a physical parcel. The paper market that eventually became the world’s most important price benchmark was, at its root, a freight-scheduling problem that got securitized.

The lesson repeats: every time the industry has been forced to reprice barrels in real time, it’s because a freight bottleneck (a closed canal, a delayed platform, a chokepoint under threat) made the old fixed-freight assumptions stop working. Crude didn’t get scarcer. Moving it got harder. The market that resulted was really a freight market wearing an oil market’s clothes.

Hormuz, again

Which brings us back to this week. Some Gulf barrels have genuinely stopped moving, this isn’t purely a shipping story, output has been curtailed too. But the point is how tightly crude price and freight cost stayed bound to each other for as long as the MOU held. Gulf countries were forced to heavily discout their barrels to be able to absorb freight cost to Asia. That’s not coincidence; it’s the same mechanism that links the WTI/Brent arbitrage to the TD25 route every trading day. The Aframax rate on that US Gulf-to-UK Continent run is a term inside the equation, the number that decides whether a US barrel actually clears in Rotterdam or stays home. Hormuz crude and Hormuz freight have been trading the same way: two prices moving as one, because the freight leg is what turns a barrel sitting in the Gulf into a barrel delivered somewhere else. Now that the ceasefire has broken, owners are pulling ships back, war-risk premiums are spiking, and every vessel still willing to transit prices in a risk premium that has little to do with the cost of producing the crude itself. Crude and freight aren’t two separate stories here. They’re one story, quoted in two currencies.

The shadow fleet is a freight market, not an oil market

If Hormuz shows freight scarcity being priced in real time, Russia’s shadow fleet shows what happens when freight capacity itself fractures into two parallel markets. Since the 2022 price cap sanctions, a shadow fleet of roughly 600–800 tankers, by some counts close to one in every six oil tankers on the water, now operates largely outside conventional Western insurance, classification, and flag. That fleet isn’t moving different oil. It’s the same Urals and ESPO crude that existed before 2022. What changed is the freight and insurance wrapper around it, and that wrapper now trades at its own price, with its own risk premium, its own enforcement risk, and its own opacity: nearly a third of tankers crossing the Baltic now sail under insurance from sanctioned Russian-linked underwriters rather than the traditional P&I clubs.

Two freight systems, one barrel. The price discount on Russian crude isn’t really a crude quality discount; it’s a freight-and-compliance discount, priced by the market’s assessment of how much harder, riskier, and less insurable it is to move that barrel versus a Brent-linked one sailing on a conventional hull.

The takeaway

Barrels are a commodity. Freight is not. You can’t substitute a stranded VLCC at anchor off Fujairah with a different one that isn’t there, and you can’t substitute a sanctioned hull’s insurance coverage with a P&I club that won’t touch it. That is why Mediterranean Aframax freight rates have quadrupled in just one week, a wave of cargoes (barrels) emerged against a tight vessel list.

Every genuinely disruptive oil-price move of the last sixty years (Suez in ‘67, the North Sea in the ‘70s, and Hormuz and the shadow fleet right now) traces back to a freight constraint more than a barrel shortage. Freight is king.

To be continued…

That slowdown has been one of the quiet factors keeping crude prices depressed over the past several months, and as the data below shows, it’s now reversing. And yet the Department of Energy’s most recent solicitation only awarded 500,000 barrels total, a function of the flattening futures curve, which matters more than it sounds, because these releases have been structured as loans, repayable with roughly 15% more volume in one to two years, not as outright sales the way the Biden-era program worked. The recent escalation in Iran has pushed the curve back into meaningful backwardation, which should in theory make selling SPR barrels attractive again. For now, though, the extra supply the market got used to over the past year is shrinking.

The Drawdown Is Decelerating, Hard

The SPR closed the week ending July 10 at 316.5 million barrels, down 3.0 million barrels (about 426,000 b/d) from the prior week. That sounds like a lot until you put it next to the pace of the week before this one, which alone ran at 881,000 b/d. In other words, last week’s release was roughly half the recent average pace, the deceleration the market has been waiting for.

Zoom out and the picture is even starker. A year ago, the SPR held 402.7 million barrels. That’s a decline of 86.2 million barrels, or 21.4%, over twelve months, an average pace across the full year of only about 237,000 b/d. The fact that recent weekly releases have been running at triple and quadruple that annualized pace tells you this hasn’t been a steady drip; it’s been a series of bursts, and last week’s number is the clearest sign yet that the burst is ending (at least for the moment).

That matters for the tape. A reserve steadily adding several hundred thousand barrels a day of supply to the market, supply that doesn’t show up as OPEC+ production and doesn’t get counted against anyone’s quota, has been a genuine, underappreciated drag on price for a while now. Take that supply away, even partially, and you remove one of the two more important bearish forces sitting quietly under the crude complex (The other one is Chinese imports).

Why the Tender Only Moved 500,000 Barrels

Here’s the part that’s easy to miss if you’re only watching the stock level. The most recent DOE solicitation for SPR barrels only awarded 500,000 barrels, a fraction of what recent draws would suggest. The proximate cause is the flattening of the futures curve, which changes the economics of the specific mechanism being used here.

This is worth being precise about, because it’s genuinely different from what happened during the Biden administration. Those releases were outright sales: barrels left the reserve, cash came in, and that was the end of the transaction. What’s happening now is structured as an exchange, effectively a loan. The barrels go out today, and the recipient is contractually obligated to return them, with roughly 15% more volume, within one to two years. It’s the same mechanism the SPR has used opportunistically for decades (crude oil loans following Hurricane Ivan in 2004 and Katrina in 2005 are the closest historical precedent), just deployed at a scale that’s actually moved the weekly stock numbers this cycle.

A loan-structured release only clears at a size the market wants when the economics work for whoever’s borrowing the barrels, and that depends heavily on the shape of the forward curve. A flat or contango curve makes borrowing oil today, using it, and repaying more later in a cheaper market relatively unattractive. A steep backwardation does the opposite: it makes today’s barrel worth meaningfully more than the barrel you’ll owe back next year, which is exactly the condition under which someone would eagerly take the DOE up on this kind of offer. The 500,000-barrel award is the market telling you the curve, until very recently, wasn’t steep enough to make the trade compelling at scale.

Iran Just Changed That Math

Which brings us to the timing problem. The Iran escalation has pushed the crude curve back into meaningful backwardation, and Brent has retraced a large portion of the selloff from $120, but it hasn’t revisited the highs above $100. The curve shape has moved more than the flat price. A market pricing near-term supply risk above deferred risk (backwardation) is precisely the setup that should, on paper, make SPR loan-exchange barrels attractive again: borrow cheap oil against a curve that pays you to hold it now rather than later. Whether the DOE’s next potential tender actually reflects that logic is the thing to watch. For the moment, the incremental barrel supply the market has been leaning on continues to shrink, not grow, even against a geopolitical backdrop that should be pulling in the opposite direction.

The Export Arb Only Just Reopened

There’s a second thread tying the price story to the flow data, and it’s been sitting quietly in the daily spot numbers. For some weeks, the WTI-Brent spread was essentially closed. A US barrel earns its way onto a tanker bound for Asia or Europe by being cheap enough, after freight, to beat a Brent-linked cargo to the same buyer. When the spread is below freight, that math doesn’t work, and the economic pull on export barrels toward the East weakens domestic crude has nowhere near as strong a reason to leave the Gulf Coast.

That changed abruptly after recent conflict escalation. Brent’s premium over WTI jumped to $7/bbl earlier this week and now sits at $5.8/bbl. So the export numbers below need to be read with that in mind: whatever pulled crude out the door last week was working against a spread that had only just started paying again.

What Else Moved

A few other data points from this week’s report worth flagging alongside the charts:

• Commercial crude fell 1.7 million barrels to 409.7 million, about 6% below the five-year seasonal average.

• Gasoline inventories fell 1.5 million barrels to 210.5 million, roughly 8% below the five-year average, a tighter print than the crude number, even with refiners running hard, which points at very strong end-demand.

• Distillate inventories built 4.6 million barrels, a genuine surprise given the direction of the other categories, though the level (108.2 million barrels) is still about 11% below the five-year average.

• Refinery utilization hit 96.2% of operable capacity on crude inputs of 17.1 million b/d, up 99,000 b/d week-on-week, refiners are running about as hard as the fleet allows heading into peak summer demand. Crack 3-2-1* at all-time highs supports this.

  *The 321 crack spread is a formula used in the oil industry to estimate an oil refinery's gross profit margin. It gets its name from the theoretical assumption that a refinery takes three barrels of crude oil and "cracks" (breaks) them down to produce two barrels of gasoline and one barrel of distillate fuel

  

• Crude exports jumped to 3.72 million b/d, up from 3.26 million the week before, but below recent highs, notable given the WTI-Brent arb spent most of the reporting window closed; worth watching whether next week’s print extends the gain now that the spread has reopened.

• Total petroleum exports (crude plus products) came in at 10.67 million b/d, even as product exports alone pulled back hard week-on-week from an unusually strong prior print.

The Setup

Put the pieces together and this week’s report reads less like a single data point and more like a hinge. The SPR drawdown that had been quietly subsidizing the market for months just cut its pace in half, at the same moment the curve flipped into backwardation and the export arb reopened after some weeks shut, three separate mechanisms, all pointing the same direction, all turning within days of each other. None of that guarantees the next DOE tender comes back big; a 500,000-barrel award is still the most recent data point we have, and the curve only just turned. But the conditions that were suppressing both the drawdown and the export flow are no longer in place, and that’s usually the part of the setup that matters more than any single week’s number.

Let me know in the comments if you find this kind of analysis useful and suscribe!.

I don’t think it’s the correct lens anymore, and the evidence for that has been piling up in plain sight for the last eighteen months. The market’s actual balancing mechanism has moved. It isn’t sitting in Riyadh. It’s sitting in Beijing, in the form of a state apparatus that has quietly become the largest discretionary buyer of crude the world has ever seen, and it is behaving exactly the way a swing producer would, just from the other side of the ledger.

This matters for anyone who owns oil, owns tankers, or is trying to figure out why the price of crude keeps refusing to do what a war in the Middle East says it should do.

What “Swing Producer” Used to Mean

The mechanics were simple. OPEC, led by Saudi Arabia, held spare capacity in reserve and used it to defend a price band. Too much oil, cut production. Too little, add it back. The 2014-16 episode and the 2020 Saudi-Russia price war were both, in their own way, proof the mechanism still worked: Riyadh chose to not play the role for a while, on purpose, and the market found out in a hurry what happens when nobody is holding the other end of the rope.

What’s changed isn’t that Saudi Arabia lost the physical capacity to swing. It’s that the marginal barrel setting the price is now more often a buying decision made in Beijing than a production decision made in Riyadh.

The Quiet Handoff: China’s 2025 Stockpiling

Through 2025, China stockpiled an estimated 0.9 to 1.0 million barrels a day of crude (Rystad Energy puts stockpiling at roughly 83% of that year’s entire import growth). This was not demand. Chinese refiners were not burning an extra million barrels a day. This was inventory management: taking advantage of a soft price environment to fill strategic and commercial storage that had sat comfortably below capacity for years.

That barrel never reached the tradeable global market. It went into a tank instead of into a ship’s discharge line, and the EIA has explicitly credited this stockpiling with supporting prices that would otherwise have fallen further. Functionally, China was doing what OPEC used to do when it cut production, removing barrels from the market to defend a price level, except it was doing it from the demand side, with a checkbook instead of a valve.

Nobody put out a press release about this. It happened quietly, month after month, in customs data most people don’t read.

The Mechanism Runs in Reverse

Then the Iran war closed the Strait of Hormuz, and the same mechanism ran backward.

Chinese seaborne crude imports fell to roughly 6 million barrels a day by June 2026, the lowest level since October 2016. Refinery run rates fell with them, particularly among the independent “teapot” refiners that had grown structurally dependent on discounted, sanctions-tolerant Iranian barrels. When that supply got cut off, a lot of that capacity simply stopped running rather than pay up for substitutes.

Here’s the part that should have gotten more attention than it did: China absorbing that shock by buying less is a meaningful part of the reason Brent never spiked to the level a full Hormuz closure should, on paper, have produced. A supply shock of that magnitude, hitting a market with normal demand elasticity, should have been genuinely ugly. It wasn’t, because the buyer on the other side pulled back exactly when the seller couldn’t deliver anyway.

China has now swung in both directions, filling up when crude was cheap, pulling back hard when it wasn’t, and the financial press has started calling it what it is. CNBC, Bloomberg, and Axios have all run pieces this year describing China as the oil market’s new swing importer. That’s not a cute turn of phrase. It’s a genuine description of who is now doing the job OPEC used to do.

OPEC’s Cohesion Is Cracking at the Same Time

The timing here is not a coincidence, and it’s not kind to OPEC’s credibility as an institution.

The UAE left OPEC entirely on May 1, 2026, to pursue production unconstrained by a quota system it had outgrown, its real capacity had drifted well above what its quota allowed for years. Iraq, OPEC’s second-largest producer, has quietly floated its own exit unless its quota, currently 4.378 million barrels a day, gets raised toward the production it says its capacity now supports. Whether or not Iraq actually walks, the fact that it’s a live conversation tells you something about how much internal discipline is left in the room.

An organization built to act as a single, coordinated swing producer doesn’t work if its second-largest member is threatening to leave over a quota dispute and its most operationally flexible member already has.

Riyadh’s Answer: The Biggest OSP Cut Since 2000

Against that backdrop, Saudi Arabia’s August Arab Light OSP to Asia landed like a statement. Aramco cut the price by $11 a barrel, the largest monthly reduction since at least 2000, pushing Arab Light to a discount against the Oman/Dubai benchmark for the first time since the 2020 price war, and to its lowest absolute level since June 2020.

You can read that as a purely mechanical response to soft Asian demand. I don’t think that’s the whole story. A cut of that size, delivered in one move, reads less like a price-taker adjusting to conditions and more like a price-setter defending share against a buyer, China, that has just demonstrated it has real leverage and isn’t afraid to use it. Whether this is the opening move of a deliberate price war or simply market-share defense that happens to look like one, it’s one more data point in the same direction: the cohesion that let OPEC act as a single swing producer for decades is coming apart at precisely the moment its actual balancing function has moved to Beijing.

The Setup

If August’s OSP cut is the opening move of something bigger rather than a one-off, the historical playbook says to expect Saudi Arabia, and now potentially the UAE and Iraq, both operating outside OPEC’s quota discipline, to add barrels rather than withhold them. That’s incremental ton-mile demand for VLCC and Suezmax tonnage on Middle East-origin routes, regardless of what price those barrels ultimately fetch.

The genuinely open question is where those barrels land. China’s own appetite is the least certain variable in this cycle, given an already-full 2025 stockpile and softening teapot-refinery runs. If Beijing stays on the sidelines, the incremental barrels have to find buyers elsewhere in Asia or Europe, which reshuffles trade routes and haul lengths more than it changes the aggregate ton-mile total. If China re-engages instead, the Middle East-to-China VLCC route (still the single largest crude trade in the world by volume) is where any rally would show up first, and a full return to China’s pre-war import run-rate of roughly 11.5-11.8 million barrels a day, nearly double the current level, would be unambiguously bullish for the entire crude tanker complex.

This comes at a time when oil-on-water, i.e. crude loaded on tankers, is at seasonal highs, which means tanker utilization is already high.

Either way, the framework that mattered for the last forty years — watch what Riyadh does, is no longer sufficient on its own. Watch what Beijing buys.

This is not investment advice. Do your own work.