Decoding Space Rocks with AI: The Meteorite Breakthrough

Artificial intelligence (AI) is transforming the way we do things, not only on Earth but also in Space.

By being used for tasks ranging from autonomous spacecraft navigation and data analysis to optimizing resource utilization and supporting scientific discoveries, the technology is enabling more efficient, autonomous, and insightful space missions.

For instance, NASA has been exploring the power of AI for many years. From autonomous rovers on Mars to AI-boosted initiatives to find new exoplanets, the agency has been leveraging this tech to enhance its understanding of space.

Recently, the US federal agency showed how AI can help orbiting spacecraft collect more targeted data. AI allowed a satellite for the very first time to foresee its orbital path, process and assess images with AI, and decide where to target an instrument, which didn’t even take two minutes or any human involvement.

“The idea is to make the spacecraft act more like a human: Instead of just seeing data, it’s thinking about what the data shows and how to respond,” said Steve Chien, a technical fellow in AI at NASA’s Jet Propulsion Laboratory (JPL) and principal investigator for the Dynamic Targeting project.

A few years ago, Elon Musk’s SpaceX also launched a satellite equipped with AI to enable the spacecraft to participate in deep space missions.

Amidst this, scientists have unraveled the secrets of the meteorite using this technology. This particular material challenges the rules of heat flow by acting as both a crystal and a glass.

With the help of AI, researchers were able to uncover the mineral’s ability to maintain constant thermal conductivity, a major breakthrough that can revolutionize materials science by transforming heat management in technology and industry. It could also help cut down the massive carbon emissions in steel production. 

How AI is Unlocking the Secrets of Meteorites

Meteoroids are the flashes of light you sometimes see streaking across the sky.

These space rocks can be as small as dust grains or as large as small asteroids. Most of them are fragments of larger bodies that have broken apart. Some come from asteroids, others from comets, and a few even come from the Moon, Mars, or other planets.

These are called meteoroids while still in space. Once they enter the atmosphere of Earth or any other planet and survive the passage, they are called meteors.

When entering the atmosphere, they do so at high speed, and as the pressure surpasses the object’s strength, it disintegrates, which leads it to burn up and give a bright flare, hence the name “shooting stars”. When they appear particularly bright, they’re called “fireballs.”

These meteors may seem like a rare occurrence, but according to NASA estimates, about 48.5 tons of such material falls on Earth every day.

Being part of space, these rocks can help provide valuable insights into the composition, formation, and history of asteroids, planets, and our solar system. 

A meteorite is made up of various materials, including rock, metal, or a combination of both.

These meteorites are studied by scientists in great detail using various techniques such as photographic and telescopic observations, radar detection, microscopy, spectroscopy, magnetometry, and others.

Lately, AI is also being used to understand space meteorites by automating their detection using drone imagery, enhancing the classification of their types through machine learning, identifying potential impact sites, and even revealing the composition of materials within meteorites. 

By analyzing vast datasets and recognizing patterns that humans might miss, AI improves the efficiency and accuracy of meteorite research, which in turn provides critical insights into the origins of life. 

For instance, research¹ from late last year found evidence of liquid water on Mars 742 million years ago with the help of a meteorite.

So, an asteroid struck Mars eleven million years ago and sent pieces of the red planet travelling through space. One of those pieces crashed into the Earth, providing us with a meteorite that can be traced directly to Mars. 

It was named the Lafayette Meteorite, and upon investigation, researchers found that while on Mars, it interacted with water. Recently, an international collaboration of scientists determined the age of the minerals in the meteorite that formed when there was liquid water. 

“We can identify meteorites by studying what minerals are present in them and the relationships between these minerals inside the meteorite.”

– Lead author Marissa Tremblay, assistant professor with the Department of Earth, Atmospheric, and Planetary Sciences (EAPS) at Purdue University

She further noted that meteorites tend to be denser than rocks on Earth, are magnetic, and contain metal. However, finding meteorites isn’t that easy. 

The chance of finding one is actually very small. As a result, researchers have been using AI along with drones to make the discovery. 

In 2022, researchers from Curtin University in Australia recovered a meteorite, one that followed an ellipse between the orbits of Jupiter and Venus, in the remote Australian outback using machine learning and two drones.

The technology allows meteorite hunters to do repetitive tasks without losing attention. In fact, the machines learn to deal with false positives through repetition. 

“The holy grail of meteorite hunting right now is a drone that can grid a geographic area, look at the ground, and find meteorites with AI.”

– Mike Hankey, The American Meteor Society

The University’s School of Earth and Planetary Sciences, along with the Paris Observatory, the International Centre for Radio Astronomy Research (ICRAR), and other institutions, meanwhile, collaborated to solve the puzzle of carbon-rich asteroids, which, while common in space, account for less than 5% of meteorites that reach Earth.

This study², which has unraveled the long-standing mystery in space science, was published this year. For this, scientists analyzed almost 8,500 meteoroid and meteorite events.

What the study has discovered is that the Sun and Earth’s atmosphere act like giant filters that destroy carbonaceous meteoroids before they reach the ground. Such meteorites are important because they contain amino acids, organic molecules, and water.

While it was already known that carbon-rich material does not survive atmospheric entry, the research showed that many meteoroids “don’t even make it that far”, breaking apart as they pass close to the Sun. 

“The ones that do survive getting cooked in space are more likely to also make it through Earth’s atmosphere.”

– Co-author Dr. Hadrien Devillepoix, Curtin’s Space Science and Technology Centre and Curtin Institute of Radio Astronomy (CIRA)

Moreover, it found that meteoroids formed by tidal disruptions are particularly fragile and almost never survive atmospheric entry. According to the Paris Observatory’s Dr. Patrick Shober:

“This finding could influence future asteroid missions, impact hazard assessments, and even theories on how Earth got its water and organic compounds to allow life to begin.”

Meanwhile, a study³ from earlier this year used AI to find that “Marsquakes”, one of the main forces shaping the planet’s surface, are caused by seismic activity from meteoroid impacts.

The team of researchers from the University of Bern and Imperial College London leveraged AI to identify new impacts in tens of thousands of orbital images of data between December 2018 and 2022 and then cross-referenced it against seismic data. It helped the researchers find 123 fresh craters to cross-reference, and out of them, 49 were a potential match with quakes.

The freshly mined data uncovered that on Mars, meteoroid impacts occur about twice as often as previously estimated.

This, Professor Tom Pike from the Imperial team noted, shows “the power of looking deeply into multiple datasets from Mars. Without the seismic data, we would not have known where to look for an impact in the orbital images, and without the orbital images, we would not have been able to locate the source of the seismic energy.”

AI has changed the game for researchers by detecting an impact in a single pixel of a low-res orbital camera that’s used for daily weather monitoring. “The power and speed of AI means we have been able to find the proverbial needle in the haystack!” he added.

The machine algorithm that played a key role here was developed at JPL, which can sift through vast amounts of data, such as images. 

AI Confirms Crystal-Glass Hybrid

Now, the latest study⁴ by scientists from Columbia Engineering has used AI to make yet another marvellous discovery. They have confirmed the “hybrid” thermal properties of a space mineral, which doesn’t follow the typical heat-flow rules. The meteorite acts as both a crystal and a glass. 

This is a breakthrough because the heat-conduction properties of crystals and glasses are completely opposite to each other. The thermal conductivities actually vary strongly in both. The thermal conductivity of materials varies dramatically depending on atomic structure. Here’s how crystalline, glassy, and hybrid materials compare:

These trends play a key role in a variety of technologies, including waste-heat recovery systems, miniaturization and efficiency of electronic devices, and the lifespan of thermal shields for aerospace applications.

Optimizing the performance and durability of materials used in these applications requires having a deep understanding of how their atomic structure and chemical composition determine the material’s capability of heat conduction.

Michele Simoncelli, assistant professor of applied physics and applied mathematics at Columbia Engineering, took the first principle approach and combined it with machine learning to identify the unique material with distinctive thermal properties.

Machine learning techniques allowed the team to overcome the computational challenges of first-principles methods and simulate atomic properties that affect heat transport with quantum-level accuracy. 

The material is the first of its kind, which was discovered in meteorites and identified on Mars. 

Figuring out the fundamental physics driving this special behavior can advance our understanding and help us design materials that manage heat during extreme temperature differences.

Now, thermal conduction, which is heat transfer through stationary matter by physical contact, depends on the atomic structure of a material. So, whether a material is glassy, with a disordered, non-crystalline structure, or crystalline, with an ordered lattice of atoms, influences how heat flows at the quantum level.

Basically, thermal conduction increases in glasses with increasing temperature and decreases in crystals upon heating.

In order to capture this opposite thermal-conductivity trend in glasses and crystals, Simoncelli, in collaboration with Francesco Mauri from Sapienza University of Rome and Nicola Marzari from the Swiss Federal Institute of Technology, derived a single equation back in 2019.

Notably, the equation describes the intermediate behavior of partially disordered materials. This includes materials used in thermal barrier coatings for heat shields, perovskite solar cells, and thermoelectrics to recover wasted heat.

Now, using this same equation, they explored the relationship between the atomic structure and the thermal conductivity in materials made from silicon dioxide (SiO2).

Also known as silica, silicon dioxide is a naturally occurring chemical compound composed of silicon and oxygen, two of the most abundant elements on Earth. It is one of the main components of sand. 

The researchers predicted that the “tridymite” form of silicon dioxide would show the signs of a crystal-glass material with a thermal conductivity that does not change with temperature. 

Tridymite is a high-temperature crystal form of silicon dioxide that occurs primarily in volcanic rocks and is formed under high-temperature and low-pressure conditions. It is also found in meteorites.

The unusual thermal-transport behavior of Tridymite prompted a team of experimentalists led by Daniele Fournier, Massimiliano Marangolo, and Etienne Balan from the Sorbonne University in Paris to run trials on a sample of silica tridymite obtained from a meteorite that landed in Germany three hundred years ago.

The experiments proved the predictions made by the researchers with measurements.

Meteoric tridymite has been confirmed to have an atomic structure falling between the orderly crystal and disordered glass. Additionally, they found its thermal conductivity to remain constant between 80 K and 380 K, the temperature range that’s experimentally accessible.

On further analysis, the team predicted that the material could form from the thermal aging in refractory bricks, which serve as a thermal barrier in furnaces for steel production.

The versatile, durable, and multi-functional steel is one of the most crucial materials in modern society, underpinning various industries and infrastructure. However, steel production is a carbon-intensive process, with just 1 kg of steel emitting about 1.3 kg of CO2. 

With almost 1 billion tons of steel produced each year, it is responsible for a lot of CO2 emissions, so much so that it accounts for about 7% of carbon emissions in the US.

As the study noted, the efficiency and environmental impact of this are largely decided by how heat is managed in furnaces, in particular through the thermal conductivity of refractory materials that can withstand extreme temperatures.

Thus, materials derived from tridymite could allow for more efficient control of the intense heat involved in steel production. So, using the study’s findings, the conductivity of refractories can be increased, in turn, reducing the burn time of furnaces and consequently lowering the steel industry’s carbon footprint.

Besides all this, Simoncelli’s group at Columbia is exploring using the same mechanisms that determine the flow of heat in hybrid crystal-glass materials to understand the behavior of other excitations in solids like spin-carrying magnons and charge-carrying electrons.

These concepts help with emerging and energy-efficient technologies, including spintronic devices, wearable devices, and neuromorphic computing.

For this, the research team is working on formulating first-principles theories to predict experimental observables, developing AI simulation techniques for quantitatively accurate predictions of materials properties, and applying them to discover and design materials to address engineering and industrial challenges.

Investing in AI Space Research

When it comes to space exploration, Lockheed Martin Corporation (LMT +0.85%) stands out for being a major contractor for NASA and the Department of Defense. The company designs AI-based satellite systems and planetary probes to support missions like Mars exploration.

The global aerospace and defense company has a market cap of $101.23 billion, with its shares currently trading at $433.60, down 11% YTD. It has an EPS (TTM) of 23.15 and a P/E (TTM) of 18.73. Lockheed pays a dividend yield of 3.04%.

Lockheed Martin Corporation (LMT +0.85%)

Just this week, the company announced its new, more capable and survivable missile warning satellite. During the testing, the Next-Gen OPIR GEO satellite proved its capability to operate in and withstand the harsh temperatures and the violent vibration conditions.

For Q2 2025, it reported sales of $18.2 billion, up from $18.1 billion in the same quarter last year. Its net earnings for the quarter were $342 million, or $1.46 per share. The company also reported $1.6 billion of program losses and $169 million of other charges. This, according to Reuters, was from “difficulties with a classified program in its Aeronautics business and international helicopter programs in its Sikorsky unit.”

During this period, cash from operations was $201 million, a massive drop from $1.9 billion in 2Q24. Meanwhile, free cash flow was $(150) million, compared to $1.5 billion in the same quarter last year. Lockheed also returned $1.3 billion to shareholders through dividends and share repurchases. 

Its CEO, Jim Taiclet, noted that US and allied customers are “asking us to elevate and accelerate many key programs,” including the US Space Force ordering additional GPS IIIF satellites. He added:

“At the same time, our ongoing program review process identified new developments that caused us to re-evaluate the financial position of a set of major legacy programs. As a result, we are taking a number of charges this quarter to address these newly identified risks.”

Latest Lockheed Martin Corporation (LMT) Stock News and Developments

Conclusion

The magic of AI is reaching beyond the boundaries of Earth to the depths of space, helping us uncover hidden patterns in space rocks, from Marsquakes to exotic thermal behaviors. With these discoveries, AI is accelerating discoveries that will transform our understanding of the universe as well as the future of materials.

Click here to learn all about investing in artificial intelligence.

References:

1. Tremblay, M.M., Mark, D.F., Barfod, D.N., Cohen, B.E., Ickert, R.B., Lee, M.R., Tomkinson, T., & Smith, C.L. Dating recent aqueous activity on Mars. Geochemical Perspectives Letters, 32, published 6 November 2024. https://doi.org/10.7185/geochemlet.2443
2. Shober, P.M., Devillepoix, H.A.R., Vaubaillon, J., et al. Perihelion history and atmospheric survival as primary drivers of the Earth’s meteorite record. Nature Astronomy, 9, 799–812 (June 2025). https://doi.org/10.1038/s41550-025-02526-6
3. Charalambous, C., Pike, W.T., Fernando, B., Wójcicka, N., Kim, D., Froment, M., Lognonné, P., Woodley, S., Ojha, L., Bickel, V.T., McNeil, J., Collins, G.S., Daubar, I.J., Horleston, A., & Banerdt, B. New impacts on Mars: Unraveling seismic propagation paths through a Cerberus Fossae impact detection. Geophysical Research Letters, first published 3 February 2025. https://doi.org/10.1029/2024GL110159
4. Simoncelli, M., Fournier, D., Marangolo, M., Balan, E., Béneut, K., Baptiste, B., Doisneau, B., Marzari, N., & Mauri, F. Temperature-invariant crystal–glass heat conduction: From meteorites to refractories. Proceedings of the National Academy of Sciences, 122(28), e2422763122 (11 July 2025). https://doi.org/10.1073/pnas.2422763122