Techno-Optimist #22
The Stargate Project launches, and so does DeepSeek. On to Mars! Asteroid mining gets real, mammoths on track for 2028, roadster mistaken for asteroid, ...and a whole lot more.
Welcome to the twenty-second edition of Techno-Optimist, your destination for all the latest updates and commentary on space, science, technology, medicine, energy, AI, and much more. Some big news, Techno-Optimist now has over 1,000 subscribers! So a big welcome to all of you, I hope you’re enjoying it here.
This newsletter is still a work in progress, and suggestions are always welcome. Today I’m tweaking a couple of categories, turning Medicine into Medicine & Biotech, and putting Agriculture by itself, at least for now. Let me know if you think of anything that pairs well with agriculture.
There’s a lot about AI in this edition. That wasn’t a deliberate choice, but rather emerged naturally from the sheer volume of news and rapidity of progress. As a result, it seems like almost every category is starting to be dominated by AI. For instance, I put the ‘AI predicts heart attacks years in advance’ under the Medicine section, but I could have just as easily put it under the AI, Engineering, etc. So buckle up, because there’s probably going to a lot more of this going forward.
Speaking of which, I had an interesting health related discussion with ChatGPT recently. Though it wasn’t like talking with a human, it was quite helpful and intelligently answered my questions, suggested new approaches, and brought up brand new (useful) information that I wouldn’t have otherwise known. My doctor later confirmed that a plan I hammered out with ChatGPT was very much along the same lines she was thinking. It’s obvious that we’re not talking about sentience here, but I do get the feeling that we might be talking about intelligence. This isn’t just a complex “mirror,” reflecting your own words and ideas back at you. Not anymore. What does that imply? I think it means that at least when it comes to silicon, intelligence can exist without sentience. I don’t think that’s possible in biology, not for complex types of intelligence anyways. It’s a strange new world.
Alright, let’s dive in.
“We will pursue our manifest destiny into the stars, launching American astronauts to plant the Stars and Stripes on the planet Mars.”
‘We can make more everything faster and better by running at "I'm the captain now" speed.’
The Stargate Project. Earlier in January, President Trump announced a $500 billion private sector AI infrastructure investment called the Stargate Project. The company will be equally funded by SoftBank, OpenAI, Oracle, and MGX, with SoftBank and OpenAI being the lead partners. OpenAI will have operational responsibility, with Arm, Microsoft, NVIDIA, Oracle, and OpenAI being the “key initial technology partners.” $100 billion will be deployed immediately, with the first data centers to be built in Texas, then in other states as well. The remaining money will be invested over 4 years, so everything should be in place by the end of 2028. OpenAI said about the project that,
This infrastructure will secure American leadership in AI, create hundreds of thousands of American jobs, and generate massive economic benefit for the entire world. This project will not only support the re-industrialization of the United States but also provide a strategic capability to protect the national security of America and its allies…All of us look forward to continuing to build and develop AI—and in particular AGI—for the benefit of all of humanity. We believe that this new step is critical on the path, and will enable creative people to figure out how to use AI to elevate humanity.
This is what acceleration looks like. If we win here and succeed in creating AGI, it’s hard to predict exactly what the resulting future will be. But rapid progress—beyond anything we’ve ever seen before in human history—is likely in everything from medicine to energy, space travel to agriculture. There’s nothing a true AGI (or an artificial superintelligence that might be built after it) wouldn’t impact or influence. The world could be changing very rapidly just a few years from now. (OpenAI) (Autism Capital) (David Shapirio)
DeepSeek. You know, when I read about the Stargate project I almost wrote how it’s a perfect example of why China can’t compete when American innovators are freed to innovate. Then DeepSeek came along. The new Chinese AI startup DeepSeek appears to have come out of stealth shortly after Project Stargate was announced. The claim was that it could do everything top American made LLMs could do, but much faster, with less energy, and 10-20x cheaper (so $5-$6 million instead of $100 million to train a new model). The stock market selloff was dramatic, with tech and energy companies taking the brunt of it, and NVIDIA in particular dropping about 16% in one day. The fear seems to be that if China can do this, aren’t all our investments in AI datacenters and the energy it takes to power them a waste both of time and money? Answer: of course not. I’ll explain.
First off, what do we know? DeepSeek’s AI does seem to be as good as, or even slightly better, than top US made LLMs like ChatGPT. Third party testing seems to have verified this, so they’re above board there. What seems far less likely to be true is the cost, with many in the tech industry calling the $5 million number “bogus,” and Elon Musk quipping that “DeepSeek obviously has ~50,000 Nvidia H100 chips that they can’t talk about due to US export controls.” The reason they might make these low model training cost claims are exactly what I said, to try and slow down investment into American AI companies, and to stifle growth in energy (now that the climate change alarmism and push for degrowth seems to not be working as well anymore).
There are some other pretty glaring issues with DeepSeek as well, the biggest being that it “reportedly collects IP, keystroke patterns, device info, and stores it in China.” There’s also a problem with censorship, it won’t criticize Chinese President Xi Jinping or communism for instance. Nor is there any point asking it what happened in Tiananmen Square in 1989.
But ignoring all that, let’s say that DeepSeek’s AI models are just as cheap as they claim. Will we need less datacenters and power plants to run our AI? Of course not. Efficiencies in energy and computing methodologies were bound to come along sooner or later, and that’s a good thing in the long run. If we can get the same AI “horsepower” for an order of magnitude less money and energy, we’ll just build an order of magnitude more AI! [sidebar: history and something called Jevons Paradox backs up my prediction. It’s the idea that as something, say energy use for a given technology, becomes more efficient, we’ll just end up using more of it. Lower cost leads to greater consumption. In the long run this is a good thing].
DeepSeek is a wakeup call for the American AI industry. This is our Sputnik moment in the great AI race. Let’s respond as those great Americans did—not by giving up, but by doubling down, accelerating even harder, and beating the Russians (or Chinese in this case) to the prize. (DeepSeek) (Palmer Luckey) (unusual_whales)
To Mars! Get in everyone, we’re going to Mars! President Trump in his inaugural address announced that Americans are going to Mars, saying “We will pursue our manifest destiny into the stars by launching American astronauts to plant the Stars and Stripes on the planet Mars.” As you can see by Elon Musk’s reaction, he was pretty pleased. While Trump will be out of office by the time astronauts land on the Red Planet, if everything goes perfectly they could launch near the very end of his term in office. He’s not talking about a flyby either, but actual boots on the ground. Buckle up, the first human to set foot on Mars could take that next giant leap for Mankind by the end of this decade.
In other Mars news, researchers have found what appear to be “two sets of ancient wave ripples on Mars,” indicating that there was water open to the air at the time they were formed. While many models of the ancient climate on Mars suggest a warm, wet past, some indicate that any water was covered by ice. It looks like this is yet more evidence that Mars was once fairly warm and Earth-like, which bodes well for soon to be explorers finding evidence of past life, or maybe even extant underground microbes. There’s only one way to really know for sure, and that’s to send humans. Robots are great, but most of the work Martian rovers have done over the decades could be done by a small team of humans in a matter of days. Once we have manned missions to Mars, the level of knowledge we have about our rusty neighbor is going to explode. (Phys.org) (Science)
Progress against cancer. There’s a lot happening on the anti-cancer front. Here’s a few of the highlights that stood out to me: Scientists have managed to trick the immune system into more aggressively attacking tumors. They did it by using lab designed viruses to “disguise” the tumors to look like pig organs, prompting an immune response. Early data from monkeys and humans indicates it’s an effective approach—slowing tumor growth, and even eliminating them in some cases.
Another approach uses specially designed bacteria to deliver anti-cancer drugs directly into tumors. The bacteria were injected into patients (just mice for now), and are cleared by the immune system everywhere except inside the tumors, where they continue growing. Then a trigger (aspirin) is given, which signals the bacteria “to invade the cancer cells and then deliver the therapy.” Human trials are slated for 2027.
A new approach for treating breast cancer has “shown the ability to eliminate small breast tumors and significantly shrink large ones in mice with a single dose.” The molecule is called ErSO-TFPy, and kills a certain type of common breast cancer without apparent side effects. Testing for safety in humans is needed, but if it works as well as it does in mice, it could be revolutionary. (Dr Singularity – cancer disguised as pig organs) (Technology Networks via Dr Singularity – bacteria) (SciTechDaily – breast cancer)
Anti-aging. AI is getting into longevity science, with a new model from OpenAI called GPT-4b suggesting ways to re-engineer something called Yamanaka factors—which can make an older cell “morph into a young-seeming stem cell.” The problem with this cellular reprogramming up until now is that it’s not very efficient, but the AI’s suggestion of editing two Yamanaka factors resulted in 50x greater effectiveness, at least in models. Another breakthrough (not AI this time) involves the development of a compound to activate an enzyme called SIRT3, “potentially reversing age related cellular decline.” Trials using it to target age related diseases—including Alzheimer’s and Parkinson’s—will begin this year. Lastly, miRNA has been used to increase lifespan in mice by reversing cellular senescence. “In middle-old aged mice (20 months of age), systemic treatment with miR-302b - delivered via exosomes - increased lifespan, with better physical and cognitive health in old age.” More work is needed on the approach to ensure that it doesn’t cause cancer, but so far it looks like a promising field of inquiry.
Speaking of age related diseases, a new study has found that the noble gas Xenon could be useful in treating Alzheimer’s disease. Inhaling the gas “reduced neuroinflammation, minimized brain atrophy, and promoted protective neuronal states in mouse models of Alzheimer’s disease.” Further work is needed to determine safety and dosage, with human trials set to begin this year. In yet another win, a team at a Spanish (Basque) university has shown that a drug called WIN55.212-2 “protects the brain and reverses the initial cognitive damage caused by dementia,” so that the rodents it was given to “behaved in the same way as those without brain damage…you could say that in some way the drug reversed the damage or protected the brain.” I really feel that in the next 5-10 years we’re going to see viable treatments for dementia. What a gift that would be. (MIT Technology Review – Yamanaka factors) (Dr Singularity – SIRT3) (Samuel Hume – miRNA) (SciTechDaily – Xenon) (Medical.net via Dr Singularity – dementia)

Fusion updates. The Experimental Advanced Superconducting Tokamak (EAST) nuclear fusion reactor in China has broken a new record, confining plasma at 100 million degrees Celsius (180 million degrees Fahrenheit) for 1,066 seconds—almost 18 minutes. This artificial sun broke the previous record of 403 seconds, which it set in 2023. It's a big step forward for China's tokamak program, and throws down the gauntlet for Western companies working to achieve commercial fusion energy first.
Also in Asia, there's three new Indian companies joining the fusion race. Pranos is taking a magnetic confinement approach, while Anubal Fusion plans to go down the inertial confinement path using lasers—an approach similar to the National Ignition Facility. The third company is working on what’s reported to be “cold fusion,” so I think we're all a bit dubious on that one.
Over in Britain, a research consortium led by the UK Atomic Energy Authority (UKAEA) has produced 5.5 tons of “reduced-activation ferritic-martensitic steel that can be used in nuclear fusion reactors.” AKA fusion steel. While it can’t stand up to the over 100 million degree plasma (nothing physical can for long), it will resist temperatures of 650 °C (1202°F) and the high rates of neutron bombardment it will encounter in the core of a working fusion power plant. (Interesting Engineering - East) (The Economic Times – Indian fusion companies) (Interesting Engineering – fusion steel)

AI and Human progress. For decades research into AI was slow, with long periods of almost total stagnation. But that all seems to be changing right in front of our eyes, with developments so rapid it’s hard to keep track of it all. And we’re only getting started. Right now in spite of how fast things are moving, we’ve only reached the knee of the curve. The real exponential change upwards is about to commence. We have CEOs of AI companies like Anthropic saying that there will be a “great acceleration” with “100 years of progress in biology over the next 5-10 years, resulting in a doubling of the human lifespan in that timeframe.” And how AI agents are on their way and will be able to “search the Tree of Knowledge to discover new and original things about the world.” [sidebar: we’re talking the ‘Tech Tree’ Tree of Knowledge, not the ‘Fall of Man’ one].
Speaking of AI Agents, OpenAI just rolled out Operator: “an agent that can use its own browser to preform tasks for you.” Currently just available to Pro users (you have to pay for that), Operator can “look at a webpage and interact with it by typing, clicking, and scrolling.” It can do tasks we normally take for granted as human only, like filling out forms, placing orders for groceries or other items, or even creating its own memes. Basically, it’s JARVIS Lite. From what I’ve heard, it’s a little rough yet, but should be improving quickly.
When it comes to education, AI will be the great equalizer (sorry human teachers). Soon we’ll have AI tutoring personalized to every student on—and eventually off—the planet. Just like how everyone, whether Elon Musk or a random guy on the street, has a smart phone with roughly similar capabilities, everyone will have access to the very best personalized AI education. A new study gives credence to this idea, finding that just 6 weeks of after school tutoring via GPT-4 was equivalent to 2 years of “typical learning gains.”
To put things in perspective, it’s important to keep in mind that as fast as things are moving now, they’ll never be this slow again. So hang on to your hats as we accelerate up the curve! (Tim Urban) (Tsarathustra – Anthropic) (Tsarathustra – Tech tree) (OpenAI – Operator 1) (OpenAI – Operator 2) (Ethan Mollick – Education)
Space

The Gaia mission is ending. Unfortunately, the spacecraft has run out of propellant, so can’t properly orient itself anymore to make observations. Fret not though, there’s a lot more data still to be released, which should keep astronomers happily busy for some time. Gaia spent over a decade making 3 trillion observations of over 2 billion stars and other objects, working to build a 3D map of our galaxy. It’s done some incredible science, and will be missed. (ESA Gaia 1) (ESA Gaia 2)

NASA’s Webb telescope has captured images of a “newfound galaxy class” galaxy that’s reshaping our views of the cosmos. Many of these “little red dot” galaxies showed evidence of containing actively growing supermassive black holes at their centers, something that hasn’t been seen before in galaxies of this age. Now that we’re able to observe them with the Webb Telescope, perhaps we’ll find this is how most galaxies start? (NASA Webb Telescope)
During the course of the space age, the size of satellites has usually decreased. But a new company called K2 is bucking that trend, and intends to use the larger launch vehicles like Starship, New Glenn, and Neutron coming into use to send up much bigger satellites. This is the way. (Ars Technica)
Asteroid mining may no longer be a distant dream. Despite some previous failures to launch as it were, Astro Forge is moving ahead with its plan to launch Odin, the first of many probes to survey and eventually sample what they hope are asteroids that can generate an economic return for the company. Dr. Phil Metzger has an excellent thread on how this could work on X. In a nutshell, he believes that if an asteroid mining company could scale up—it would bring their costs down. The end result would “bring the cost to market down well below Earth’s best mines,” because the concentrations of platinum and other valuable metals are so much higher in some asteroids than on Earth. There’s a lot of risks, and no guarantee of success, but it’s encouraging to see Astro Forge forging ahead. (AstroForge, via Dr. Phil Metzger)
New research suggests that star collisions were key to the formation of the universe’s first supermassive black holes, which the JWST has observed as early as 450 million years after the Big Bang—far earlier than expected. The study suggests that supermassive black hole ‘seeds’ of a few thousand solar massed “could form rapidly in dense and dynamic star clusters in the early Universe.” The nascent supermassive black holes could then grow rapidly by mergers of these seeds, explaining the observations. Future more sensitive gravitational wave detectors might be able to find evidence of these mergers, validating the hypothesis. (SciTechDaily)
AI, Energy, Engineering, & Physics
Meta just dropped a major translation breakthrough—their new machine learning system can instantly translate written speech between 101 languages and spoken words in 36 languages. If you're thinking "this sounds like the Babel fish from Hitchhiker's Guide, or the Universal Translator from Star Trek," you're not alone. What's particularly impressive is that it can translate speech-to-speech directly without needing to convert to text first, and the delay is just a few seconds—on par with human translators. It’s also open source for non-commercial use. (Nature)
Mayo Clinic just announced a big shift towards AI, partnering with Microsoft Research and Cerebras Systems to speed up diagnoses and improve accuracy. The Microsoft collaboration focuses on AI equipped radiology—they're building models that can automatically generate reports from chest X-rays, check if tubes and lines are placed correctly, and spot changes from previous scans. Pretty impressive stuff. The Cerebras partnership involved creating a genomic foundation model that combines reference genome data with Mayo's patient exome data (the part of our genome where most disease-causing mutations happen). They're already seeing promising results with rheumatoid arthritis, helping predict which treatments will work best for specific patients—instead of the usual months long trial and error approach. (Mayo Clinic) (Cerebras)

Microsoft researchers have introduced MatterGen, a generative AI model designed to streamline the discovery of new materials for specific applications, like more efficient batteries, better heat tiles for re-entering spacecraft, or perhaps even room temperature superconductors down the road? Unlike traditional trial and error or computational screening methods (both ‘needle in a haystack’ approaches), MatterGen directly generates novel materials based on user defined design requirements, such as chemical, electronic, and mechanical properties. It “enables a new paradigm of generative AI-assisted materials design that allows for efficient exploration of materials, going beyond the limited set of known ones.” MatterGen could revolutionize materials design, and I’m excited to see what it does this year. (Microsoft Research, via @IterIntellectus)


Two different developments have led to some very interesting potential improvements to body armor. The first is a team from Northwestern University that has “developed the first two-dimensional (2D) mechanically interlocking material.” Resembling chain mail, it “exhibits exceptional flexibility and strength,” recommending it for better, lighter body armor. The second is based on work out of Caltech, where interlocking material called PAM (Polycatenated Architeched Materials) respond “to some stresses like a fluid and to others like a solid.” These PAMs don’t exist in nature, and can have almost endless variation, allowing them to be fine tuned for different applications, anything from packaging or soft robotics, to body armor or medical devices. (Phys.org) (Caltech)
A team of Korean researchers has developed a miniature computing chip that mimics the brain’s ability to learn and correct errors. Designed for neuromorphic computing, the chip can automatically separate moving objects from backgrounds in video streams, improving performance over time. This breakthrough not only demonstrates the commercial viability of next-gen neuromorphic systems but also promises to boost everyday device performance by enabling AI processing locally. With enhanced privacy, faster speeds, and better energy efficiency, this chip could reshape how we handle AI tasks on devices, and reduce reliance on cloud servers. (Dr Singularity)
Exoskeletons are generally used to help people where they have to lift heavy loads, or for increasing walking/running endurance outdoors. But this new exoskeleton aims to help piano players improve by moving “individual fingers to guide complex motions.” This early version looks a little unwieldy, but it actually improved pianists’ skill and speed with just half an hour of use. (Interesting engineering)

I talk a lot about AI, but there’s another realm of computing that’s moving ahead: supercomputers. The El Capitan supercomputer is now the fastest in the world, with a top speed of 2.746 exaFLOPS. That’s a 1 with 18 zeros behind it, operations per second. It’s a million times faster than teraFLOPS. (Live Science)
Medicine & Biotech

American company Science Corp. is creating a unique brain implant that uses artificial neurons to fuse and communicate with biological ones. The company believes it's implant will be less invasive than current technology like Neuralink, and "will be more general-purpose and reach many areas of the brain, as opposed to current implants that target specific regions like the motor cortex or visual cortex." If successful, it could eventually be used on humans too, allowing restoration of brain function lost to disease or injury, and even instantly learning new skills. We’ll see if it works. (Ashlee Vance)
Neuralink’s brain-computer interface has been successfully implanted in a third patient, and it seems that the company will be picking up the pace in 2025, with plans to do 20-30 more implants by the end of the year. (@unusual_whales)
One of the biggest problems with using drugs like Ozempic for weight loss is that loss of muscle often accompanies it—a very undesirable outcome. A study from California’s Salk Institute has found that “a protein called BCL6 is key to maintaining healthy muscle mass.” Increasing it reverses muscle losses in mice, suggesting that a drug boosting BCL6 could be paired with GLP-1 targeting weight loss medications so that fat—but not muscle—gets the boot. (Salk Institute)
Scientists have designed AI generated proteins that neutralize deadly snake venom toxins, potentially revolutionizing antivenom treatments. Unlike traditional plasma derived antivenoms, these proteins are cheaper to produce, more effective against elusive toxins, and easier to manufacture at scale. In tests, they provided near total protection for mice exposed to lethal doses of cobra venom. While not a full replacement yet, this breakthrough could enhance existing treatments and pave the way for newer, low cost therapies—offering new hope for millions affected by venomous bites worldwide each year. (SciTechDaily)
Chinese researchers have managed to reserve some of the damage in mice caused by Parkinson’s Disease by "simulating dopamine-producing brain cells wirelessly with gold nanoparticles." The nanoparticles are coated with antibodies that guide them to their targets in the brain, and are then activated by "near-infrared light shone through the skull." The light gets converted into heat, triggering cellular changes "that promote repair while releasing peptides that help break down and clear out the alpha-synuclein protein fibrils clogging up Parkinson's brains." (Science Alert)
A paralyzed man just piloted a virtual drone using only his mind. Thanks to a brain implant, patient T5—who had suffered a spinal cord injury—controlled the drone in real time by imagining finger movements. The implant decoded his brain signals with unprecedented accuracy, allowing him to maneuver the quadcopter as if using a video game controller. The breakthrough, part of the BrainGate2 trial, goes beyond piloting drones. By refining mind-controlled movement, it could enhance digital access, robotic prosthetics, and even multiplayer gaming for people with paralysis—bringing brain machine interfaces closer to everyday life for many more people. (Singularity Hub)
AI powered CT scan analysis could predict heart attacks years in advance. Researchers used AI to analyze calcium scoring CT scans, uncovering hidden cardiovascular risk factors beyond arterial plaque. The model integrates heart shape, visceral fat, and other biomarkers to improve risk prediction and timing of heart attacks. With $4 million in NIH funding, this approach could revolutionize preventative care—allowing earlier interventions, reducing healthcare costs, and ultimately saving lives. (Interesting Engineering)
Scientists have managed to engineer immune evasive CRISPR enzymes, potentially making gene editing safer and more effective. They redesigned Cas9 and Cas12 nucleases to avoid immune detection while maintaining high editing efficiency. By identifying and modifying immune triggering protein sequences, they significantly reduced immune responses in mice. This breakthrough could improve the safety and longevity of CRISPR based therapies, expanding their potential for treating genetic diseases. (Phys.org, via Dr Singularity)
One of the biggest challenges in bioprinting organs is getting adequate vascularization of the tissue. A new approach uses a 4D bioink to enable precise control over blood vessel formation in 3D-printed tissues. Using aptamers—"small DNA structures that can bind biochemical signals and release them in a targeted manner”—the ink mimics the body's natural vascularization process by releasing growth factors on demand. This breakthrough could solve a major challenge in bioprinting by ensuring oxygen and nutrient delivery, bringing functional, transplantable tissues closer to reality.(3D Printer Magazine)
GMOs improve human lives. Couldn’t put it more starkly than this. (Simon Maechling)
Agriculture
It’s not something most of us think about, but out of the hundreds of thousands of plant species, only 15 provide ~90% of our calories, despite tens of thousands being edible. So what makes plants good targets for domestication by humans? Plasticity—how adaptable a plant is. This makes it easier for us to manipulate. Genetic make up—simpler is better. If traits humans want to change are controlled by a small number of genes, that’s better for us. Mutation rate—faster is better. While we have domesticated plenty of trees, none are amongst the top calorie producers for humans, likely because of their long lifespans and relatively slow mutation rates. By studying this, researchers are hoping to add to our basket of domesticated plants, both entirely new, and maybe cousins of plants we already eat. (Phys.org)
John Deere has announced that its machines will now be equipped with Starlink, allowing for “seamless farm connectivity.” It doesn’t matter how remote they are, as long as they can see the sky they’ll always be online, allowing farmers to improve operational efficiency, particularly during planting and harvest. (Farm Progress)
Weird & Wonderful
Work to bring back mammoths just got a boost, with Colossal Biosciences raising $200 million in a Series C funding round. The company says they are on track to “produce a woolly mammoth calf born to a surrogate elephant mother by late 2028.” Colossal also plans to resurrect the dodo, and Tasmanian tiger. (Tech Central) (Colossal Biosciences)
It’s not often a new species is found in a photograph, but that’s exactly what happened here. A new dinosaur was identified from pictures of bones from a German museum, where they were destroyed in 1944 during a bombing raid. It’s quite a distinct animal, estimated to be 10m (33 feet) long, making it one of the largest therapod (carnivorous) dinosaurs known. It’s been named Tameryraptor markgrafi, or "thief from the beloved land,” and lived in Egypt around 95 million years ago. (Live Science)
An old forest of 5900 year old Whitebark pine has been uncovered due to a melting glacier in the Rocky Mountains of Wyoming. Whitebark pine doesn’t grow there now, it’s too cold for it. Meaning that almost 6000 years ago—during the Holocene Climate Optimum—temperatures were warmer than today. Maybe we’ll survive this whole global warming thing after all. (New Scientist, via Chris Martz)
Polyfluoroalkyl substances (PFAS) are usually referred to as ‘forever chemicals’ because they don’t readily break down, and persist over long periods of time in the environment, and in our bodies—where they can cause health problems. A team from the University of Buffalo may have found a way to help deal with the problem: a type of bacteria that “can break down and transform at least three types of PFAS, and, perhaps even more crucially, some of the toxic byproducts of the bond-breaking process.” The group now plans to work on the bacteria to make it more efficient at breaking PFAS and their biproducts down. (Science Daily)
Always double check your asteroids. Earlier in January, an amateur astronomer identified what was thought to be a new asteroid. But upon checking, astronomers at the Minor Planet Center for Astrophysics realized it wasn’t a rock at all. Turns out, it was Elon Musk’s Tesla Roadster, that he launched into space on the maiden voyage of SpaceX’s Falcon Heavy rocket back in 2018. (Smithsonian Magazine)
Photos & Videos




Here’s a really stunning series of photographs showing the same event from the perspective of Earth looking up, and from space looking down. It’s a collaboration “12 years in the making” between NASA astronaut Don Pettit, and Babak Tafreshi from National Geographic. (Don Pettit + Babak Tafreshi, via A. Pettit) (Don Pettit)
You may not be interested in space, but space is interested in YOU! For what was probably the first time ever, a meteorite has been captured by video and audio as it hits the ground in Prince Edward Island, Canada. Watch carefully though, it’s moving fast so blink and you’ll miss it. (IFL Science)
Two cells checking out what’s going on nearby, curtesy of Nikon Small World photography. (Jennifer Silverman, Nikon Small World Photography)
NASA’s James Webb Space Telescope gives us a beautiful look at the fine details of a supernova remnant Cassiopeia A. You can see the intricate patterns of dust lit up in the infrared, looking like gnarled old tree bark, or perhaps the mineral patterns in a metamorphic rock. (NASA)

Just because the JWST gets a lot of the glory these days doesn’t mean the old, faithful Hubble Space Telescope doesn’t have more surprises still. Have a look at this huge photomosaic of the Andromeda Galaxy, the largest ever assembled from Hubble data. (NASA)


Comet C/2024 G3 (ATLAS) has broken up now, but wowed skywatchers in the southern hemisphere before it did, being clearly visible to the naked eye. (Yuri Beletsky) (NASA)
Andrew McCarthy has been a frequent source of excellent photographs that I’ve had the privilege of sharing here. Here’s a 3 minute video compilation of some of his favorites (click on the photo to see the video on X). (Andrew McCarthy)
Recommendations & Reviews

Today I’m recommending you all go read an article by Tomas Pueyo, who writes Uncharted Territories here on Substack. It’s called ‘Final: So Where Should We Build Ten New Cities in the US?’ It builds off a series of 8 articles he’s written—links are in this one, half are free to read, the other half paywalled—that dives into more detail about the proposed cities. This final chapter gives a list of all 10, and a short blurb on each.
Personally, I think it’s a great idea. Building some or all of these cities would help usher in a new era of excitement and dynamism to the US, and therefore to the world. Maybe it could help rev up a new Roaring Twenties? (Uncharted Territories)
That’s it for now, but Techno-Optimist will be back in two weeks. Thank you all for reading—and until next time, keep your eyes on the horizon.
-Owen















So many things to get excited about this week; I was all set to declare Asteroid mining as my fave, not to bring down prices to get platinum here on Earth, but to eventually build things in space from the materials, but then you had to go and recommend Uncharted Territories, so that has to be first now. Bring on the New Cities!
PS. A topic that goes with Agriculture is Food more generally
"Let me know if you think of anything that pairs well with agriculture." Pottery! A writing system! The development of an elite class of priest-kings!