A solar panel the size of a drink coaster split water into hydrogen at 10.8 percent efficiency two years ago, and when the same company built one twelve times bigger the number dropped to 9, which is the part nobody put next to this month’s announcement

Ten percent is the number direct solar hydrogen has been chasing since before most people had heard of green hydrogen. Hit 10% solar-to-hydrogen efficiency in a panel you can actually manufacture, and the cost models start producing a number that competes with buying solar panels and an electrolyzer separately. Miss it, and you have a very expensive science fair project.

SunHydrogen said on August 11 that its modules cleared the line in an Australian partner’s lab, and that both companies are moving to pilot testing in South Australia.

The claim is real and both sides call it preliminary. What nobody put in the headline is the part where the efficiency went down the last time these panels got bigger.

Two years ago the same modules hit 10.8%, then lost almost two points

Here is the track record, which matters more than any single lab result.

In October 2024, SunHydrogen reported that its 100-square-centimeter modules, built with German manufacturer CTF Solar, reached 10.8% solar-to-hydrogen efficiency in testing at Honda R&D in Japan. Solid number. Small panel, roughly the size of a drink coaster.

Then the company scaled up. It built a 1-square-meter array and had 1,200-square-centimeter modules tested at Professor Kazunari Domen’s lab at the University of Tokyo, which is about as authoritative an address as this field has.

Those modules came in at 9% active-area efficiency. SunHydrogen called it the highest reported figure for a module that size, which may well be true. It is also 1.8 points below what the coaster-sized version managed.

That is the whole problem with this technology in one comparison. Efficiency at bench scale does not survive the trip to panel scale intact, and panel scale does not survive the trip outdoors intact either.

The new result comes from somebody else’s lab, which is the point

Every efficiency figure SunHydrogen has published came from SunHydrogen or a company it pays or partners with. This one came from Sparc Hydrogen, which had no obligation to say anything flattering.

Sparc’s testing put the modules above 10% and found they produced more hydrogen as the light was concentrated, which is the behavior its reactor is built around. SunHydrogen CEO Tim Young called it “exactly the kind of outside validation our shareholders want to see.”

That framing is worth sitting with for a second. SunHydrogen is a nine-person company in Coralville, Iowa, with no revenue, roughly 5.3 billion shares outstanding and a stock trading around two cents. It was incorporated in 2009 as HyperSolar and renamed in 2020. Seventeen years of announcements, no product.

None of which makes the science wrong. It does mean the company has strong reasons to publicize a lab number, and readers should weight it accordingly.

Fortescue is in this because it already quit the other approach

Sparc Hydrogen is a joint venture between Sparc Technologies, the University of Adelaide and Fortescue, through a wholly owned subsidiary. Its reactor uses photocatalytic water splitting, invented by University of Adelaide chemistry professor Greg Metha, which produces hydrogen from concentrated sunlight and water with no electrolyzer and no grid connection at all.

Andrew Forrest’s Fortescue Future Industries first took a stake in 2022 and moved to an equal top ownership position in January 2025. Michael Dolan, Fortescue’s head of R&D, sits on the board.

The timing tells you what this is. Fortescue gutted its own green hydrogen program in 2024, scrapping a target of 15 million tons a year by 2030 and cutting 700 jobs, because electrolysis powered by renewables cost too much at that scale. Then it doubled down on a technology that skips electrolysis entirely.

That is a hedge, and a rational one. It is also not a vote of confidence in green hydrogen broadly.

The local backdrop is worse. South Australia canceled a $600 million hydrogen hub planned for Whyalla, which would have included the world’s largest electrolyzer, and moved the money to rescuing the Whyalla steelworks instead.

SHARP is real hardware, and it has been running for a year

The Sparc Hydrogen Advanced Research Pilot sits at the University of Adelaide’s Roseworthy campus north of the city. It came online in June 2025, and Sparc believes it is the only facility anywhere purpose-built to test photocatalytic water splitting under concentrated solar conditions.

Stage one prototype testing happened back in late 2023 at CSIRO’s Energy Centre in Newcastle, using 451 heliostats aimed at a 100-foot solar tower.

In December 2025, Sparc announced sustained hydrogen generation at Roseworthy, completing commissioning. The plant currently runs on photocatalyst powders from Shinshu University in Japan.

SunHydrogen’s modules would go into the same reactor as a different chemistry. That is the actual experiment: does a photoelectrochemical panel work inside a rig designed for powders?

The fine print splits the market in half

Most coverage of this deal has called it a partnership with off-ramps. The actual terms are more specific than that.

Sparc gets rights to apply SunHydrogen’s technology in concentrated-light applications above an agreed threshold. SunHydrogen keeps everything in decentralized production under natural, unconcentrated sunlight. That is a field-of-use split, not a shared bet.

If the collaboration finishes successfully, Sparc holds an 18-month option to negotiate a long-term supply agreement or a manufacturing license, plus a right of first offer against competing arrangements in concentrated light. Each side keeps its own IP and its own improvements. The whole thing runs 24 months with independent review points and go or no-go decisions between phases.

And Sparc Technologies told the Australian Securities Exchange the agreement is expected to have limited near-term financial impact. When the listed partner tells its own shareholders not to get excited, that is the most honest sentence in the announcement.

The Austin pilot broke before it got fixed

SunHydrogen’s other program is at the Hydrogen ProtoHub at the University of Texas at Austin, and it has not been smooth.

Earlier prototype modules suffered a voltage drop in the solar substrates and coating failures that let moisture reach the semiconductor layer. Those were fixed at lab scale with help from CTF Solar, and the pilot restarted with upgraded modules.

Worth remembering when reading efficiency numbers. Moisture reaching a semiconductor stack is exactly the failure mode you would expect from a device whose entire operating principle involves sitting in water, and concentrated sunlight makes thermal stress worse rather than better. Panels lose performance as they heat up, which is why desert solar installations fight temperature so hard.

SunHydrogen also has a roughly $2.3 million order with CTF Solar for 1,000 modules of nearly 2 square meters each, plus new offices in Japan and Austria. As of this spring the company reported about $33 million in cash and no debt, which buys time but not a product.

The number that settles this is not efficiency

Two things have to land.

The modules need to hold above 10% bolted into the SHARP reactor at Roseworthy, outdoors, for months rather than hours. Given that the 1,200-square-centimeter version already gave up almost two points indoors, that is not a formality.

Then the jointly funded techno-economic assessment has to produce a levelized cost of hydrogen that beats solar plus an electrolyzer in a sunny place. That is the deliverable both companies named, and it is the only number that decides anything. Efficiency is a talking point. Dollars per kilogram at the fence line is a business.

The useful thing about a 24-month agreement with review gates is that it produces an answer either way. Either the panels make hydrogen under an Australian sun at a price that works, or Sparc walks at a gate and keeps its powders. Hydrogen programs rarely fail loudly. They go quiet, and the next press release is about something else.