Hormuz isn't empty. It went dark

Every open chart of the Strait of Hormuz says it is closed. It is not. Since the US and Israel struck Iran on February 28, 2026, the ships that free satellite data can see crossing the strait fell to almost nothing. Yet the oil kept moving: most of it now crosses without showing up in open tracking data, under US naval escort. The strait did not go empty. It went dark, and that changes who you should believe.

Dark, not empty

In the last week of September, open satellite data saw 7 tankers cross Hormuz. Kpler counted 72 million barrels of crude going through.

September 27 to October 3, 2026. IMF PortWatch (ships with tracking on); Kpler via CNBC.

3%

Ships seen crossing Hormuz in open data, vs before the war

76%

Crude still getting through, vs before the war (Kpler)

+42%

Frontline, a supertanker owner, from the eve of the war to publication

1. What the open data sees

ships seen that day

▶

Source: IMF PortWatch, daily chokepoint transits from AIS data. Events: Wikipedia, 2025–2026 Iran–United States negotiations.

On February 27, satellites still saw 64 ships cross. On March 4, for the first time, they saw none. Watch the peace deals, though. After the Islamabad Memorandum in June, the visible ships came back, up to 51 on June 24. When the ceasefire broke on July 8, they vanished again within days. Real cargo does not appear and disappear that fast; transponders do. The open count measures how safe captains feel to be seen, not how much oil moves.

2. What actually gets through

Sources: IMF PortWatch (AIS transits, week to October 3, vs January 1 to February 27, 2026); Kpler and Windward via CNBC, October 6, 2026.

In the last week of September, PortWatch saw seven tankers cross, and together they were smaller than one supertanker. In the same week, Kpler, a commercial tracker, counted 10.3 million barrels a day of crude through Hormuz, about 23% below its prewar level. That is roughly 36 supertanker loads a week that the open data does not show. Windward, another tracker, puts the flow a bit lower, but in the same range. The oil goes along a route next to Oman’s coast that the US Navy protects, often transferred to other ships in the Gulf of Oman to keep tankers out of reach.

3. Who to believe

Sources: JODI Oil World Database (crude exports reported by Kuwait); IMF PortWatch (tanker capacity crossing Hormuz, AIS).

Commercial trackers sell their data, so they have a reason to see what nobody else sees. I wanted a check that does not depend on them. Kuwait is one: it has no pipeline around the strait, so every barrel it exports crosses Hormuz, and its government reports those exports every month. In April and May, Kuwait’s exports fell to almost zero and it cut production to under a quarter. The open data agreed: the strait really was shut that spring. Then the two lines split. By July, Kuwait was back to 82% of its prewar exports, while the tanker capacity PortWatch saw stayed at 13%.

And a government that publishes its own collapse is not dressing up its numbers. The blind spot is also older than the war: from November 2025 to January 2026, PortWatch counted a third fewer ships crossing Hormuz, while Kuwait kept exporting and Jebel Ali in Dubai, a port ships can only reach through Hormuz, stayed as busy as before. My first draft of this study read that dip as ships fleeing ahead of the war. It was not.

4. Who pays, and who gets paid

Frontline FRO

At publication

$53.75

Oct 5, 2026 close

→

Now

…

Loading live price

Difference since publication

…

Brent fund BNO

At publication

$61.45

Oct 5, 2026 close

→

Now

…

Loading live price

Difference since publication

…

Source: Yahoo Finance daily closes. Live data refreshes on page load.

The blockade works one way: in September, Iran loaded no crude at all, while its neighbours’ oil kept flowing. Moving oil in the dark is expensive. Shipping a tanker of Gulf crude to China costs about $1 million a day, and in the third quarter Iran attacked about two ships for every hundred that crossed. Since July, at least nine sailors have died. Somebody earns that risk premium: Frontline, one of the largest owners of supertankers, is up 42% since the eve of the war. BNO, a fund that tracks Brent crude, is up 77%. Both are frozen at the last close before this post and update live every time the page opens.

So what

I see two lessons. First, open data is going blind exactly where it matters most: the places where ships have reasons to hide. A free dashboard showing “Hormuz traffic −97%” is right about what it counts and wrong about the world. Second, the market already prices this. Oil is high not because the strait is closed but because getting through it is slow, dangerous and costly. As Lloyd’s List’s editor Richard Meade put it, “The oil market is not becoming more secure. It is becoming more efficient at operating under sustained insecurity.” That can last as long as the escort does.

What the numbers leave out

• Two kinds of count. PortWatch counts ships whose tracking transponder (AIS) is on. Kpler and Windward combine AIS with satellite imagery, port data and other sources they do not publish. We compare each source with its own prewar level, not the raw numbers with each other.
• Four published figures. The commercial numbers are a weekly Kpler average and a Windward range, as reported by CNBC. We do not have their daily series, so the chart shows one recent week, not a history.
• Ships, not barrels, in the open data. A count does not show cargo, which is why the chart also uses PortWatch’s tanker capacity (deadweight tons).
• Government figures are self-reported. JODI publishes what each country submits and marks these figures as not yet assessed. Kuwait’s production is flat at exactly 2.58 million barrels a day from December to February, which looks like an estimate. The latest month is July 2026, so the check does not cover August and September. Customs data of the buyers (Japan, Korea, India) would be the next independent check.
• Why the blind spot started in winter. We do not know. GPS jamming in the Gulf and ships switching transponders off are the usual reasons; Windward recorded jamming hitting more than 1,100 vessels in a single day once the war began.
• Schematic map. The animation is a drawing, not a map: shapes, lanes and dot positions are illustrative. Only the number of ships a day comes from the data.
• Share prices are not profits. Frontline’s shares react to freight rates everywhere, not only in the Gulf.

Data

Download the datasets: hormuz_daily.csv (ships a day through Hormuz by type, with capacity, 2019 to October 4, 2026) and chokepoints_monthly.csv (Hormuz, Suez, Bab el-Mandeb and the Cape of Good Hope, monthly, vs their 2025 average). Source: IMF PortWatch.

Methodology

• Ships seen: IMF PortWatch, daily chokepoint transits built from satellite AIS data, January 1, 2019 to October 4, 2026. “Before the war” is January 1 to February 27, 2026; the comparison week is September 27 to October 3, 2026, the same week as the Kpler figure. Tankers and tanker capacity (deadweight tons) are PortWatch’s tanker category. Used under the IMF terms, with attribution.
• Crude that gets through: Kpler (10.3 million barrels a day in the week ended October 3, vs a prewar 13.5 million) and Windward (9–10 million vs 14.5 million; we use 9.5), both as reported by CNBC on October 6, 2026. The same article is the source for the escorted route along Oman, ship-to-ship transfers, the $1 million a day freight cost, the Q3 attack rate (Windward), the sailors’ deaths (International Maritime Organization) and the Richard Meade quote.
• Supertanker loads: a very large crude carrier (VLCC) carries about 2 million barrels and is 160,000 to 320,000 deadweight tons, per the EIA. 10.3 million barrels a day for seven days is 72 million barrels, about 36 loads. The seven tankers PortWatch saw that week totalled about 123,000 deadweight tons.
• Blind spot: Hormuz transits averaged 92 a day in July–October 2025 and 60 in November 2025–January 2026. Over the same periods, Jebel Ali averaged 25 and 24 port calls a day, from PortWatch’s daily port data. Jamming: Windward, March 1, 2026.
• Government check: Kuwait’s monthly crude oil exports and production, in thousand barrels a day, from the JODI Oil World Database (Joint Organisations Data Initiative), July 2025 to July 2026. Both Kuwait’s exports and PortWatch’s tanker capacity are shown as a share of their own July 2025 to February 2026 average.
• Iran’s loadings: zero in September 2026 (Bloomberg, Kpler, Vortexa; early satellite reads), via Briefs and Scott Bessent.
• Events: dates from Wikipedia’s 2025–2026 Iran–United States negotiations.
• Share prices: Yahoo Finance daily closes. “Since the eve of the war” is the close on February 27, 2026 vs the close on October 5, 2026, the last before this post.
• Animation: each day, ships appear at a rate proportional to that day’s PortWatch count (tankers in proportion to that day’s tanker share), split at random between the two directions. The speed is eight days a second.
• Code: plain Python scripts (collect.py, analyze.py, snapshot_stocks.py).
• Trademarks: names of companies, funds and data providers belong to their owners. This site is independent and not affiliated with or sponsored by any of them. This study was produced with Claude, a model made by Anthropic.