It is not down on any map. True places never are.

Herman Melville, Moby Dick

Intro

Harvard studies proved space tech investors collectively have lost money. Yet, investors pour millions of dollars into companies that promise to find minerals from space, and some of the most sophisticated venture capitalists in the world, the people who funded Google, who backed Tesla's supply chain, who manage the retirement savings of Canadian schoolteachers, have decided that watching the ground from space is worth paying for

Why are the smartest minds in venture capital pouring money into an industry with historically terrible financial metrics?

Remote Sensing

History of remote sensing has de-risked the operational side of going to space, and the technology itself has proved its value.

The root of remote sensing in mineral exploration is in the air, and it began with an oil company in California in 1920. Union Oil geologists assembled 400 aerial photographs, taken from a company airplane with an ordinary camera, into a mosaic representing 6,250 acres over the Santa Fe Springs and Richfield prospects. That approximated a coherent picture of the land below. And it worked; the method helped geologists read the surface topography of the land in ways that walking across it could not. The oil discoveries at Santa Fe Springs came in 1923, and the images were part of the story.

The success of Union oil is still the reason investors pay for Earth monitoring tech, more on that later. But exactly 100 years later, investors paid for the same concept but from orbit.

Western Mining Corporation understood the useful part when, in December 1933, they flew two de Havilland Dragon aircraft over the Eastern Goldfields of Western Australia, looking for gold. They were the first private company to attempt systematic aerial photography for mineral discovery at a regional scale, and the effort failed. A camera, even a very good one, pointed down from an aircraft, can only tell you what the surface looks like. It cannot tell you what the surface is made of, and it certainly cannot tell you what is hidden under it.

After the Cold War ended, the United States declassified more than 800,000 images taken by spy satellites between 1959 and 1972, photographs of extraordinary resolution for their time. It is tempting to think of what these images might reveal about the geological evidence on minerals. But the cameras that took these photographs were not, in any meaningful sense, mineral detection instruments. 99% of companies raising funds from 2015 attempted to fix this.

This distinction between the shape of things and the chemistry of things is the central problem that every remote sensing technology since 1920 has been trying to solve, with varying degrees of success and at varying distances from the ground.

What Does a $100 Million Camera See?

A normal camera captures light across three broad bands of the visible spectrum. Red, green, and blue. This is because the human eye has three types of color receptors, and a camera designed to produce images that look natural to a human eye needs only to approximate what those receptors respond to. The result is a photograph. It tells you about colour, texture, and shape. It does not tell you about chemistry.

Minerals, however, are far beyond what the eye can see. Every mineral absorbs and reflects specific wavelengths of electromagnetic radiation in a pattern as distinctive as a fingerprint, and most of the wavelengths that carry the richest mineralogical information sit in the infrared.

What could be useful, and did not exist until the 1970s, was a camera that could see all of these hundreds of narrow slices of the electromagnetic spectrum. Which brings us back to the original question, the one that has not yet been answered: why do investors pay for this?

Bipolar Era Who Won the Hyperspectral War

The Risk That Was Already Retired

The Gemini missions conducted jointly with the U.S. The Geological Survey of the mid-1960s is easy to overread. But the cameras on Gemini were ordinary film cameras, producing ordinary photographs, and the USGS geologists who studied the images were looking at the same three channels of visible light that Union Oil's pilots had captured over California in 1920, only from much higher up and across a much larger area. The fundamental limitation did not.

Gemini implication for investment in 2015 were:

first, that putting a camera in orbit and returning usable images of the Earth's surface was operationally possible,  that the engineering problem of getting an instrument into space and pointing it at the ground was solved, or solvable, decades before anyone had a commercial reason to do it. For investors in 2026 writing checks to Pixxel or Orbital Sidekick or Wyvern, this matters: the operational risk of satellite imaging was retired not by any of the companies they are funding, but by a government program that flew sixty years ago.

The second thing Gemini proved, less intentionally, was that ordinary photography from orbit was not sufficient for mineral detection. The images were useful for reading large-scale geological structure.

This is the thread that connects Gemini to Landsat to every hyperspectral startup that has raised venture capital in the last decade. What was needed was a different kind of instrument entirely.

The Instrument That Changed Everything. And Still Wasn't Enough

Landsat, launched in 1972, was the attempt to build one. It was not hyperspectral in the modern sense, collecting hundreds of narrow bands.This was enough to begin separating rock types that looked identical in visible light.

What happened after Landsat is where the story becomes genuinely contested, and where the competitive advantage of the twenty-first-century companies is rooted. Technology divided the field. On one side were the researchers and geologists who saw in multispectral and eventually hyperspectral imaging the beginnings of a new kind of mineral detection. On the other side were the practitioners who had spent careers finding mines and were unpersuaded that the view from space added much to what a skilled geologist with good maps and ground-truth samples could already do.

The Ambiguity That Made It Fundable

The argument has never been fully resolved. The discovery of the Collahuasi and Ujina copper deposits in northern Chile, among the largest copper discoveries of the late twentieth century, involved Landsat 5 imagery in the identification of surface alteration zones, which is often cited as proof of the technology's value. But Masterman, a respected geologist, in his analysis of what actually found the world's greatest copper deposits, argues more carefully: the multispectral data was not the discovery mechanism. It was a validation layer, a way of confirming and refining targets that experienced geologists had already identified through conventional means. According to their reading, you did not need the satellite to find the copper.

Why Speed Is Worth More Than Discovery

This distinction is everything to an investor. A technology that discovers minerals is a different business from a technology that validates mineral targets.

Geologists optimize for the process, tech founders for the outcome, and investors for the outcome in the shortest possible time.

Validation does not sound like a billion-dollar business. But the mandate today driving the money is electrification. It requires copper, lithium, cobalt, nickel, and rare earths in quantities the existing mining industry cannot deliver fast enough. Investors are not paying for discovery only but for speed of discovery. A technology that shrinks a five-year exploration program to eighteen months (even theoretically) is worth a great deal when the clock is the constraint.

The Moment Private Capital Took the Leap

Until 2015 several attempts were made to launch multi-use satellites, some successful but discontinued. In 2015 private capital took the first leap, indirectly.

The IoT Company That Accidentally Became the Most Important Business in Mining

Fleet Space Technologies was an IoT connectivity company. Its architecture was built around the problem of sending small amounts of data from remote sensors in places with no cellular coverage to the internet, via a constellation of small, cheap nanosatellites in low Earth orbit. The business was connecting things in the middle of nowhere to everything else. The infrastructure that makes a remote mine site manageable has nothing to do with finding the mine in the first place.

The same architecture is the backbone of what made mineral exploration possible in 2023.

Blackbird Ventures backed Fleet Space in 2017 because the architecture matched their mandate: early frontier technology with the potential for global distribution. What Blackbird was paying for was the position Fleet would occupy by the time anyone else tried to follow. In 2017 Fleet and Blackbird didn’t have any plan for mineral exploration.

Their theoretical moat:

You can outspend technology. You cannot build overnight the relationships with physical asset owners that come from years of devices running reliably in their most remote operations. By the time a competitor could replicate the satellite constellation and the ground hardware, Fleet had distribution in forty countries, multi-year contracts with asset owners, and satellites already in orbit. The moat was the accumulated friction of getting there first.

The Pivot

But distribution without margin is not economically viable. IoT data was becoming a commodity, and commodity businesses do not raise Series B rounds at meaningful valuations. Something had to change. This pivot rewrote the history of mineral exploration and venture capital.

Here is where speculation is honest and worth making. Blackbird had also backed Earth AI, an AI company active in mineral exploration. Whether Blackbird connected the two companies directly is unknown. What is visible is the outcome: Fleet pivoted toward mineral exploration, and the architecture they had built for IoT connectivity turned out to be precisely what subsurface sensing required. Sensors in remote areas, transmitting small amounts of data to satellites, are processed and returned as actionable intelligence. The product changed, but the infrastructure did not.

The Investor Who Wasn't Supposed to Be There

Grok Ventures joined at Series B. Grok's stated mandate is climate and clean technology not mineral extraction/exploration, which makes their investment look counterintuitive until you read it correctly. What Fleet's architecture offered was a way to process subsurface physics at the edge. The system didn't scan the ground from orbit; instead, ground-based seismic sensors called Geodes listened to the earth's ambient vibrations non-invasively, utilizing Fleet's nanosatellite constellation purely as a secure, remote data pipeline to transmit those massive geophysics signals to space. For Grok, the investment case was not the mineral, but the method: a hybrid space-and-ground stack that could locate critical minerals faster and with less environmental destruction than anything that existed.

The competitive advantage of Fleet before the Series B:

Orbital slots, spectrum licenses, and regulatory approvals are finite. The fleet already held them. That is not something capital alone can replicate.

By early 2023, before the Series C closed, Fleet had a working satellite constellation, a proprietary AI platform, and contracts active in minimum 30 countries with Rio Tinto and Barrick. Three things that, together, no competitor possessed. But they also hold data before the disturbance, a state no competitor can replicate without going back in time. They capture the data, miners scorch the ground and move on. It’s worth mentioning the parallel of the Fleet’s strategy with Terra AI. Both were outsiders to mining, developed a tech and locked themselves inside the incumbent.

Whether you could have predicted this pivot from competitive intelligence alone, looking from outside in 2021, is a fair question. Three signals were visible to anyone paying attention.

Three Signals

The first signal was financial. Fleet was in a low-margin IoT business, and they were not alone in discovering this. Maxar and Planet Labs had already seen the same wall and were pivoting away from raw data toward analytics and applications. A new fundraising round without a new direction would have been difficult to justify to any serious investor. The announcement of the Series B was the signal.

The second signal was the portfolio. Blackbird had backed Earth AI. Venture capital is a network of business portfolio companies that find each other. Whether the connection between Fleet's remote sensing infrastructure and Earth AI's low-impact drilling was made explicitly or arrived at independently is unknowable from the outside. But the proximity was visible to anyone reading the portfolio carefully.

The third signal was In-Q-Tel. The CIA's investment arm has one mandate: find commercially viable technologies with dual-use value for US intelligence and national security. They do not invest in commodity IoT businesses for agricultural sensors in Australia. When In-Q-Tel backed Fleet Space in late 2021, they were looking for secure, tactical data transmission capabilities in denied or remote environments. But the genius of the play was what came next: the exact same satellite architecture built to relay battlefield telemetry was perfectly positioned to handle the massive data streams required for subsurface geophysics. The question of why a US intelligence vehicle would back an Australian company sat unresolved until you looked at what else In-Q-Tel was funding at the same time.

The Intelligence Hiding in Plain Sight

Orbital Sidekick started in a garage in San Francisco. Their first raise was small enough that the investors remain unidentified in any public record, which usually means the money came with something more valuable attached,  intellectual property. In 2018 the next investor was Allied Minds, which only backs commercially viable intellectual property. Then In-Q-Tel. The pattern is consistent: someone had something worth protecting before anyone had a product worth selling.

What Orbital Sidekick was building was hyperspectral satellite imaging for energy infrastructure monitoring, the 1920 oil reconnaissance problem, restated in the language of 1970s multispectral technology, deployed from orbit.

In-Q-Tel invested in both Fleet Space and Orbital Sidekick. One company listening through the ground. One company reading the surface from above. The subsurface layer and the surface layer of the same intelligence stack, funded by the same organisation.

In 2012, the National Intelligence Council had published its Global Trends 2030 report, identifying critical mineral scarcity as a structural threat to the energy transition and therefore to US economic and national security. The question of who would fund the solution to that shortage is answered, at least partially, by looking at where In-Q-Tel placed its bets in the decade that followed.

Why OSK was a defensible company?

OSK’s moat comes from multi-year defence contracts, their imaging resolution is slightly lower than Pixxel (a sensing company). The most important is they never contributed to a real discovery as profound as Fleet Space. Inside energy sector and defense has some achievements though.

The Crowded Room Nobody Has Won

Since 2019, private capital backed a wide range of space tech targeting energy, ESG and mineral explorations. Almost all the companies offer a multi-use satellite in different flavors of hyperspectral imaging. The hardware itself is becoming a commodity, some companies like OSK and Pixxel tried to build a moat by creating data to intelligence stack that ends in 2026 by proprietary AI analytic platforms. The rest, still try to figure it out.

Who Will Survive?

Even inside the crowded market of satellite imaging, no one has the substantial share of the market, no company has the superior technology, no permanent moat has ever been built and there is a lot of room to implement new technologies.

None of the companies yet take equity in mines. AI analytics is defensible in 2026 to 2030, but the data sensing companies gather are not disappearing irreversibly. The data is publicly available. The Fleet business model seems very defensible until someone with a superior technology takes equity in the mines. The survivors is the fastest company that acquires data that can be captured only once.

The survivor also needs to focus, no dedicated venture exists to focus only and only on mineral exploration, even Fleet Space is a multiuse entity.

There is something almost paradoxical in the final accounting. A Harvard analysis of every venture-backed space company in the United States found that the industry's Sharpe ratio sits at a dismal 0.09,  far below the threshold of 1.0 considered the minimum justification for the risk taken, meaning that for every dollar of excess return these companies generate, investors absorb roughly eleven dollars of variance they are not being compensated for. SaaS space companies do even worse, returning negative IRR at the Series A stage on average. By any conventional financial measure, the money should not be flowing. And yet in 2021 alone, $15 billion went into space companies, shattering the previous record. The explanation is that the investors bet on what an energy transition requires, also intelligence-adjacent vehicles whose mandate is explicitly non-commercial are not optimising for Sharpe ratios. They are optimising for a different kind of return, one the Harvard model does not capture.

The National Intelligence Council said it plainly in 2012: the public sector would identify the problem of critical mineral scarcity but could not move fast enough to solve it. Private capital read that forecast and concluded that the gap between what governments could see and what they could do was itself the market. The money flowing into space tech for mineral exploration is the price of speed, paid by people who cannot afford to wait for the public sector to catch up.

Back in 2012, the world had not collectively agreed to net-zero targets, and global electrification was not yet an absolute economic mandate. This massive pressure to build a clean energy infrastructure is exactly why private investors are willing to completely disregard traditional financial metrics. They understand that finding these minerals just ten years earlier means ten fewer years of burning coal and relying on aging grids, saving a massive volume of greenhouse gas emissions. The capital isn't chasing standard venture returns; it is funding a desperate, subsidized race against time to secure the foundational components of the next century's industrial economy.

So far, companies like Fleet Space, Orbital Sidekick, Wyvern, and Pixxel have taken very narrow, specific approaches to solving this crisis. But hyperspectral imaging and Ambient Noise Tomography (ANT) are far from the only ways to map the earth's mineral wealth. There remains a vast amount of unallocated space in orbit, unique physical sensor combinations that haven't been tried, and deep commercial moats waiting to be built by anyone who can connect orbit to geology. The door is wide open, but as resource nationalism tightens globally, it will not stay open for long.

If you are an investor looking to accurately navigate this frontier without falling into the "tough tech" traps outlined by Harvard, or a founder trying to secure finite orbital licenses and build lasting asset relationships, you need a strategy that looks beyond the hardware hype. Let's connect.

References

COMPANY PRESS RELEASES / PRIMARY CORPORATE SOURCES

Allied Minds plc. (2018, April 9). Allied Minds leads funding round at Orbital Sidekick [Regulatory News Service filing]. London Stock Exchange. https://alm.ir-data.com/article/id/129

Allied Minds plc. (2019). Annual report and accounts: For the year ended 31 December 2018. London Stock Exchange. https://www.alliedminds.com

Allied Minds plc. (n.d.). About us. http://www.alliedminds.com/about-us-2/

Fleet Space Technologies. (2021, November 17). Fleet Space Technologies secures $26.4m (USD) in Series B funding to enable global satellite connectivity for Internet of Things. https://www.fleetspace.com/newsroom/fleet-space-technologies-secures-26-4-m-usd-in-series-b-funding

Fleet Space Technologies. (2024, December 12). Fleet Space closes A$150M Series D with A$800M+ valuation. https://www.fleetspace.com/newsroom/fleet-space-closes-a-150m-series-d-with-a-800m-valuation

Fleet Space Technologies. (2025, January 15). Fleet Space's exploration satellite network expands with next-gen Centauri satellites. https://www.fleetspace.com/newsroom/fleet-spaces-exploration-satellite-network-expands-with-next-gen-centauri-satellites

TRADE & SPECIALIST JOURNALISM

Albon, C. (2023, July). Orbital Sidekick: Taking hyperspectral imaging from the garage to the Pentagon. Breaking Defense. https://breakingdefense.com/2023/07/orbital-sidekick-taking-hyperspectral-imaging-from-the-garage-to-the-pentagon/

Australian Financial Review. (2023, November 13). This moonshot tech firm is Australia's fastest-growing company. https://www.afr.com/work-and-careers/management/this-moonshot-tech-firm-is-australia-s-fastest-growing-company-20231113-p5ejhe

Coldewey, D. (2023, June 1). Pixxel's hyperspectral orbital imagery attracts Google cash. TechCrunch. https://techcrunch.com/2023/06/01/pixxels-hyperspectral-orbital-imagery-attracts-google-cash/

Roberts, R. (2025, February 23). 2024 Global Mining Technology Awards: The winners. InvestMETS. https://www.investmets.com/2024-global-mining-technology-awards-the-winners/

Thorn, A. (n.d.). Boost for hyperspectral satellite monitoring as Orbital Sidekick raises $16M. Optics.org. https://optics.org/news/boost-for-hyperspectral-satellite-monitoring-as-orbital-sidekick-raises-16m

GOVERNMENT & INSTITUTIONAL SOURCES

Grok Ventures. (n.d.). Portfolio. https://grok.ventures/portfolio/

In-Q-Tel. (n.d.). Portfolio [Orbital Sidekick entry]. https://www.iqt.org/portfolio?2d34173a_page=7&c6fb1237_page=14&company-name=orbi#full-portfolio

NASA. (n.d.). Landsat 1. https://science.nasa.gov/mission/landsat-1/

National Intelligence Council. (2012). Global Trends 2030: Alternative worlds. Office of the Director of National Intelligence. https://www.dni.gov/files/documents/GlobalTrends_2030.pdf

The Aerospace Corporation. (n.d.). A brief history of space exploration. https://aerospace.org/article/brief-history-space-exploration

U.S. General Services Administration. (n.d.). EO-1 Hyperion [Data set]. Data.gov. https://catalog.data.gov/dataset/eo-1-hyperion

U.S. Geological Survey. (n.d.). Earth Observing-1 (EO-1). https://www.usgs.gov/centers/eros/science/earth-observing-1-eo-1

PEER-REVIEWED ACADEMIC SOURCES & BOOKS

Chenevey, E. B. (2023). Houston, we have profits: Analyzing venture capital investment in the space technology industry [Bachelor's thesis, Harvard University]. Harvard DASH. https://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37378272

Haldar, S. K. (2018). Essentials of mineral exploration and evaluation. Elsevier.

Halbouty, M. T. (1976). Application of LANDSAT imagery to petroleum and mineral exploration. AAPG Bulletin, 60(5), 745–793. https://doi.org/10.1306/83D92754-16C7-11D7-8645000102C1865D

Masterman, G. J., Cooke, D. R., & Moore, R. L. (2005). Geology and discovery of porphyry Cu-Mo-Ag deposits in the Collahuasi District, Northern Chile. In T. M. Porter (Ed.), Super porphyry copper & gold deposits: A global perspective (Vol. 1, pp. 175–188). PGC Publishing.

Sabins, F. (1999). Remote sensing for mineral exploration. Ore Geology Reviews, 14(3), 157–183. https://doi.org/10.1016/S0169-1368(99)00007-4

Welty, E. M., & Taylor, F. J. (1966). The 76 bonanza: The fabulous life and times of the Union Oil Company of California. Lane Magazine & Book Company.