Carbon credits from a plankton bloom: how the Planktos business model came apart

In the first week of May 2007, a small Californian company called Planktos was preparing to spread iron over a patch of the eastern Pacific west of the Galápagos Islands, grow a plankton bloom, and sell the carbon dioxide that bloom was supposed to absorb as offsets. The plan never went ahead in that form. Within a year it had been shelved, and the company soon wound down.

Most retellings treat Planktos as a geoengineering story, which it was. It is also a compact case study in something more ordinary: how to build a business on a product nobody can see, weigh or inspect. That side of the story has aged worse than the science, because the market Planktos relied on is still with us.

The product was a certificate

The science fits in a few lines. In large stretches of ocean, plankton are short of iron, so adding a little can set off a bloom. Some of the carbon those tiny plants take up sinks when they die or are eaten, and carbon that reaches the deep ocean can stay out of the atmosphere for a long time.

But Planktos was not selling plankton, or iron. Its revenue was meant to come from the step after the bloom: an estimate of the carbon dioxide removed, converted into tonnes, then into certificates, then sold to people and companies wanting to cancel out their own emissions. The ship and the iron were costs. The certificate was the product. (Carbon accounting trips people up at the first step: a tonne of carbon becomes about 3.67 tonnes of carbon dioxide, a conversion our CO2e calculator handles along with methane and the other gases.)

So the economics hung on a single ratio. Iron is cheap. If a modest load could lock away a very large tonnage of carbon, each tonne could be sold cheaply and still leave a healthy margin. Anything that shrank that leverage (a smaller bloom, less sinking, a faster return of carbon to the surface) came straight out of the gross margin.

Who was going to pay

Ocean fertilization had no way into the regulated carbon markets of the day. The Kyoto Protocol's Clean Development Mechanism had no methodology for it, and Europe's emissions trading system would not take it. That left the voluntary market: travelers offsetting flights, companies making carbon-neutral claims, institutions after good press.

In 2007 that market was small, growing fast and loosely governed, worth by industry estimates a few hundred million dollars a year. No single standard dominated; the Voluntary Carbon Standard, later the trade's largest certifier, launched only in November 2007. A seller could, in effect, write its own method, hire its own verifier and set its own price. For an unproven technique, that looseness was the attraction, and the mood of 2007 was receptive to climate pitches of every kind. Even patent filings that year promised microbes engineered to make hydrogen or ethanol.

Planktos worked the land side of the trade too. In mid-2007 a Hungarian affiliate, KlimaFa, offered the Vatican a reforestation project in Hungary, publicized as a way to make the city-state carbon neutral. Reporters who checked in later years found little sign the trees had been planted. A donated forest was cheap marketing for a firm that needed buyers to trust its tonnes.

Then there were investors. Planktos was raising money from shareholders as well as courting offset buyers, so the voyage was a demonstration for both. A visible bloom and a news cycle would help the next funding round whether or not anyone could show where the carbon went.

What a buyer was really assuming

Strip away the branding and a Planktos tonne was a stack of assumptions, all of which had to hold.

Here is the arithmetic the sales material left out, with numbers invented to show its shape. Give each of the first five rows a generous 80 percent chance of holding, and the chance that all five hold at once is 0.8 to the fifth power: about one in three. Partial failures are likelier than total ones, so the real discount is a matter of degree. But uncertainties multiply, and a seller who prices every tonne as certain keeps the difference.

Measurement made it worse. A bloom drifts with the currents, and the carbon sinking out of it has to be estimated from sediment traps, chemical tracers and models by specialist teams at sea. Every dollar spent doing that properly raises the cost of a tonne sold as cheap. Skimping on measurement looks more profitable on paper.

Title was murky too. A bloom in international waters has no owner, and no state had agreed to host, approve or register the credits.

How the story was told

Early coverage largely followed the company's framing. A New York Times feature in the first days of May 2007, picked up by British papers, presented the venture as an enterprising experiment, with oceanographers' doubts in a supporting role. Planktos cast the trip as a voyage of recovery and leaned on the Galápagos link to Darwin. For a business selling a belief about invisible carbon, that attention was close to the whole marketing budget.

Critics used the same channels. The technology watchdog ETC Group issued a news release that week arguing that the science was turning against the idea, and the dispute spilled onto environmental blogs, where the company hit back and offered public debates. Ecuadorian groups objected to any experiment near a World Heritage site; the company said its spot was in international waters, far from the islands.

The science turned that spring

In late April 2007 a French-led team reported in Nature on the Kerguelen Plateau in the Southern Ocean, where iron arrives naturally and steadily. That slow supply moved carbon to depth far more efficiently than the short, artificial doses of earlier experiments, and the authors cautioned against reading their results as support for commercial fertilization. In business terms, the leverage behind the pitch had just been marked down.

On May 4, the Intergovernmental Panel on Climate Change released the summary of the mitigation volume of its Fourth Assessment Report. It called geoengineering ideas such as ocean fertilization largely speculative and unproven, with a risk of unknown side effects, and noted that no reliable cost estimates had been published.

In June the scientific groups of the London Convention and London Protocol, the treaties governing dumping at sea, issued a statement of concern: too little was known about the effectiveness and impacts of iron fertilization to justify large-scale operations. The parties endorsed it in November. The United States had brought the Planktos plan to their attention, and the Environmental Protection Agency had told the company that a dump from a US-flagged ship would need an ocean-dumping permit. As reported at the time, Planktos said it would use a vessel registered elsewhere, an early lesson in regulatory arbitrage at sea.

In January 2008 a group of prominent ocean scientists argued in Science that it was premature to sell offsets from commercial-scale fertilization until it had been shown to remove carbon, keep it out of the atmosphere and have acceptable, predictable effects. For a certificate seller that was worse than a ban. A ban closes a market; a scientific consensus that your product may not exist spoils it for the very buyers who were paying to say the tonne was real.

Running out of road

Its ship, the Weatherbird II, finally left Florida in late 2007, the Galápagos plan dropped and talk turning to the Atlantic. In February 2008 Planktos announced that it was postponing its ocean work indefinitely, blaming a campaign against it and an inability to raise the money it needed. Within months the company had wound down.

Rivals fared no better. Climos, a San Francisco start-up that presented itself as the careful, science-led alternative, and Australia's Ocean Nourishment Corporation, which wanted to use urea rather than iron, never sold ocean-fertilization credits at any meaningful scale. The formal rules came later in 2008, when parties to the Convention on Biological Diversity and the London treaties moved to rule out fertilization other than legitimate research. The money had left before the rules arrived.

What the episode still prices in

Seventeen years on, the voluntary market is far larger and more organized, with registries, rating agencies and, since 2023, principles from new integrity bodies on what a high-quality credit should look like. It also spent 2023 absorbing a media investigation into whether rainforest credits from the biggest standard represented real cuts, part of the uneven effort to finance nature by measuring it that a piece here on the road to Rio+20 traces.

Ocean carbon is back on the order books too. Since 2022, coalitions of large companies have committed money to buy durable carbon removal in advance, and several marine start-ups working on alkalinity, sunken seaweed and similar ideas have sold tonnes ahead of any widely agreed way of measuring them. Most are far more careful than Planktos was. The structure of the trade is the same.

Three lessons from 2007 still apply. First, the product is the measurement, so whoever writes the methodology captures much of the value. Readers of this blog's look back at the 2007 seed rankings will know the pattern from another industry, where control runs through patents and licenses rather than sales. Second, uncertainty has to be priced, not footnoted; a tonne resting on seven assumptions is not worth a tonne. Third, when a field trial is also a sales event, the pressure to report a big number is built in from the start.

The question to put to any new climate fix sold through offsets is the one the early coverage of Planktos skipped. Whether the bloom will happen is the easy part. What matters is who will count what it did, and who is paying them to count.