AI x Deep Tech Market Review (March 2024)

This post was originally a 15m “future of AI” presentation delivered to the Mindstone AI community on March 6, 2024.

I could have called this The Age of Autonomy or The Age of Acceleration, or perhaps something else will stick, but I went for the SEO juice instead. Nevertheless, it does seem like a new era is kicking off.

I also thought about calling this “Drones vs Clones,” like some old school campy sci-fi movie:

But that actually sounds a bit bleak, which brings us to the first theme today.

Theme 1: Science Fact & Science Fiction

Joe Lonsdale recently said to e/acc founder Guillverme Verdon (aka @BasedBeffJezos) on his American Optimist podcast:

Many have pointed out that sci-fi has gotten more dystopian, and that we’re making less of it overall recently. Maybe we don’t possess the imagination or intellect to see beyond what’s upon us, possibly the fastest technological transformation in human history. On the other hand, the future is what we make it. We must be proactive and first-principled in where we steer it.

The Singularity is Near Here

It’s worth noting that famed futurist Ray Kurzweil and author of the book “The Singularity is Near” predicted in 1999 that we would achieve ‘human-level AI’ by the 2020s. We seem on our way.

What most don’t remember is the book’s subtitle: “When Humans Transcend Biology”. His most iconic claim: the Singularity, when humans and machines will have merged so profoundly that we will be something new and evolved, is predicted for 2045.

That same year, Gen Alpha, today’s elementary school kids and younger, will range between 18 and 30 years olds. That’s nuts. Ray Kurzweil thinks the Singularity will happen when my son is 25 and my daughter is 23. Hell, I’ll only be 58, I guess. But damn, what a crazy time to grow up!

I’m fascinated, inspired, and yes a little bit worried about the world my kids and family will live through over the next 5, 10 and 20 years. If you are too, then welcome to the show. :)

With that lens in mind, and in the spirit of getting our science fact and science fiction game on, here are some inspiring books I recommend:

AI 2041: Ten Visions for Our Future (sci-fi/non-fiction blend)

• An AI expert teams up with an author to publish 10 sci-fi short stories projecting the impact and role of AI by the year 2041, accompanied by brief non-fiction analysis behind each story to help ‘break it down’.

• Kai-Fu Lee: Former President of Google China, senior exec at Microsoft and Apple.

• Chen Qiufan: Award-winning author, translator, creative producer, and curator. President of the World Chinese Science Fiction Association.

Nexus Trilogy (sci-fi), More than Human (non-fiction)

• Author Ramez Naam, former Microsoft computer scientist for 13 years.

• Nexus: A consumable ‘drug’ that permanently implants a nanodevice in your brain stem, enabling users to ‘connect’ to the internet through a brain-to-computer interface ‘using just their thoughts’… and from there users can ‘connect’ with one another, telepathically.

• More Than Human: Several non-fiction essays by Naam exploring the potential for transhuman applications with chapter titles like Designer Bodies, Designer Minds, and even Designer Children.

Wild.

Life 3.0: Being Human in The Age of AI (sci fi short story intro, non-fiction)

• Max Tegmark: MIT Physicist & Computer Scientist

  • Life 3.0 is when we upgrade our hardware.

• The Tale of the Omega Team: Developed a specialized AI extraordinary at one task: programming AI systems. Great sci fi short story!

The Coming Wave (non-fiction)

• Mustafa Suleyman: Cofounder of Deepmind, Inflection AI (makers of Pi)

• Thesis: 2 huge waves coming at once — AI + Synthetic Biology.

• Other major complementary waves:

  • Energy ← fusion, fission, batteries, superconductors, solar

  • Robots/Drones ← “Embodied AI”

  • Quantum ← Computing taken to limits of physics

Theme 2: AI x Deep Tech

Deep tech: Projects based on breakthroughs in science and technology (my definition, they vary).

• Alt definition: Deep tech is anywhere there’s more technical risk than market risk.

• Deep tech does not have to be AI at all, but this intersection will be our focus because this is where the greatest acceleration is happening in our burgeoning new tech era.

I want this channel to forecast as vividly as possible where the world could go 5, 10 and 20 years from now, and there’s no better window into that future today than AI x Deep Tech.

Why AI x Deep Tech?

• Unique datasets — AI companies need unique datasets in order to differentiate, and cutting-edge science and technology fields have highly specialized, continuously updating datasets.

• Generative AI — Models can be trained not only to produce poems and ad copy, but also 3d simulated chemicals and organics, unlocking massive improvements to traditional trial-and-error scientific method.

• Autonomy — Certain deep tech processes may only be possible with, or capital efficient at scale, if they are run by autonomous machines (ie AI robots).

This market is highly dynamic and pretty broad, but I am tracking 6 major categories at the moment (plus a surprise 7th I’ll reveal at the end!):

• Computing

• Material Science

• Robots/Drones

• Space/Defense

• Ag/Food/Climate

• Life Science

Let’s take a stroll through these disciplines to see how AI is accelerating innovation and bringing the future forward.

Computing

Models

• ChatGPT (OpenAI, Microsoft), Claude (Anthropic, Amazon), Llama (Meta), Mistral (Microsoft), Grok (X)

• “Compute coming online is increasing 10x every 6 months.” – Elon Musk. 

  • That is 5,000x faster acceleration than Moore’s Law, doubling every two years.

• Linear relationship (so far) between amount of training data and performance, meaning no diminishing return (yet)… and we’ve already thrown the whole internet at it. (h/t Intro to LLMs by Andrey Karpathy)

Chips

• Hardware and material science will also continue to innovate, accelerating what models we can train and interact with.

• Watchlist includes: Nvidia, Groq, Atomic Semiconductors

Quantum / Thermodynamics

• Watchlist includes: Rigetti, Quantinuum, Extropic

• Extropic CEO and founder Guillverme Verdon is also the e/acc founder.

• He’s a former quantum computer scientist taking a fundamentally different approach. From their recently released litepaper:

Well, damn.

Material science

The study of innovating computing hardware is ultimately one of material science, as evidenced by some of the examples above, but AI and material science go way beyond just computer chips.

Material discovery 

By training models on the periodic table, molecular chemistry, and physics, scientists are able to product ‘GenMat’ AI models.

• MatterGen by Microsoft: Generative model that enables broad property-guided materials design

• Citrine Catalyst: Your Digital Assistant for Materials and Chemicals Research, Development, and Deployment

Strong areas of interest include:

• Silicon and computer chips

• Carbon and carbon capture

• Metals for electricity/energy (Lithium), construction/manufacturing, robotics

• Ceramics for nuclear fusion/fission

• Polymers for plastic replacements, biomaterials (implantables)

3D printing 

Advances in robotics, material discovery, and generative designs will accelerate scale in industrial, manufacturing, and space use cases.

Energy 

Nuclear, solar, batteries and superconductors are all material science projects accelerating faster thanks to AI.

Plus, we just discovered the world’s largest lithium deposit in California, so we are set! 💪

Robots/Drones

“Embodied AI”

• Self-training, autonomous robots with natural language interfaces.

• Self-learning neural nets accelerating their speed of training and level of safe autonomy.

• Watchlist includes:

  • Figure, Tesla ← leaders in ‘consumer-facing’ humanoid robots, though their uses will be wider than that.

  • Boston Dynamics ← Spot helps the police, inspects industrial parks, and does dangerous jobs so people don’t have to.

  • Agility Robotics ← Amazon’s warehouse robotics partner.

    • Demo video: Digit + Large Language Model = Embodied Artificial Intelligence

  • Dexterity: “AI-Powered Robots That Transform Warehouses”

Autonomous Machines

• Outrider: “Autonomous yard operations for supply chain and logistics”

• Reshape: Automated microscope analysis via Petri dish computer vision.

• Simbe: Autonomous retail store robot that dispatches staff where most needed in real-time while also answering customer questions.

Full-Self Driving (FSD)

Not just Tesla, not just cars.

• Saildrone (drone ship), Bedrock (drone submersible)

• Whisper Aero (electric jet), Hermeus (drone airplanes, including for consumers).

Now we’ll build on these fundamental categories — computing, material science, and robotics — with 3 more application-specific categories:

Space / Defense

Will also include ‘Security’ or ‘Safety’ here as well, such as tools for law enforcement and home defense.

Rockets / Satellites

• Neuraspace: AI-powered satellite traffic manager

• Relativity Space: 3d printed rockets

Space Factories

• Varda: “Manufactures unique products in microgravity and returns them for use on Earth via reentry capsules.”

AI Weapons & Deterrents

• Shield AI: “Building the World’s Best AI Pilot” ← pretty terrifying, actually

• Aerodome: Autonomously deployed aerial drone support for law enforcement. ← also creepy

• Ambient.ai — AI-powered security cameras that report suspicious activity as its happening using computer vision

• Prepared — AI assistant for 911 operators and callers

• Flock — AI-powered neighborhood watch

Ag / Food / Climate

Crop Yield Optimization

• farm-ng: “a robot for every crop, task, and terrain.”

  • “Bringing AI to the Field” ← Great copywriting! 🙂 

• Halter: “Virtual fencing for high-performance grazing”

  • It’s an AI-assisted collar for cows that beeps them into place based on real-time GPS. No more farm dogs! 😮

• Shinkei: “As little as 1 in 3 fish make it to a plate.” Their autonomous “harvest suite” reduces waste caused by humans.

Synthetic Nutrients & Materials

Of course, if you can use Gen AI to discover new inorganic chemicals and materials, no reason the same tech can’t be used for organic substances, too.

• Ohalo Genetics: “Our technology shrinks the breeding cycles of plants. Our speed, accuracy, and scalability create entirely new varieties of the world’s most loved and consumed crops”

  • Shout out to All-In Bestie David Friedberg, Ohalo CEO

• Modern Meadow: Creators of a synthetic leather more real than ever before.

Sustainability

• AMP: “The first zero-manual-sortation materials recovery facility (MRF)”

  • They built a robot the size of a room or bigger that can automatically sort and bundle recycling heaps for reuse. Bad ass!

• Built Robotics: “Robots for the Solar Revolution”

  • They modified an ‘off the shelf’ excavator into an autonomous pile-driver capable of installing 300 steel rods in one day while a full human crew can manage only 100. Built is focused specifically on bringing down cost and time on major solar farm buildouts.

• Rainmaker: “Uses large-scale numerical modeling of cloud formations, rapidly deployable drones, and new chemicals in order to seed clouds precisely and maximize precipitation.”

  • AI + Drones + Material Science = Rain

Life Science

What’s especially notable about this category in my research thus far is how much money these companies have raised, included below:

Drug Discovery

Same as material science, there are many researchers, companies and institutions developing foundational models based on genomic data, protein structure, and other areas of life science in order to generate potential new therapies.

• Tempus ($1.5b funding): “AI-enabled precision medicine”

• Insitro: “Through the power of machine learning and data at scale, we decode the complexities of biology to unlock transformative new medicines.”

• Valo: “What if we could accelerate the creation of medicines on behalf of patients?”

• Generate Biomedicines: “A generative model for protein design.”

• Together AI: “Long-context biological model that generalizes across DNA, RNA, and proteins, capable of prediction tasks and generative design.”

Neuromodulation

Brain waves are getting decoded. While some are going straight to implantations (we’ll get to that next), others are doing wearables instead. The devices detect the presence or absence of specific brain wave patterns to determine whether you’re tired, rested, in the flow, etc., and then take some action to either reinforce that feeling or try to change it.

• Elemind: “Wearable neurotechnology that augments sleep, attention, and ultimately the human experience.”

• Prophetic: “Non-invasive neurotech device to induce and stabilize lucid dreams.”

And now for our surprise 7th category!

Transhuman

There is some consensus among futurists that humans and machines will increasingly merge over the next 5, 10 and 20 years, potentially leading to something altogether new for our species.

• Orchid: “Have healthy babies with the world’s first whole genome embryo screening test.”

  • CEO Noor Siddiqui believes IVF and embryo gene optimization will be the default way to have babies in the not-too-distant future. Listen to her awesome podcast with genetics pioneer George Church here.

• Atom Limbs: “The first artificial human arm.”

  

  Wow. The iPhone for arms.

  • Robot-grade advanced prosthetics

  • “We’re making artificial arms that will be capable of near-full human range of motion, restore a basic sense of touch, and be mind-controlled.” ← Is it weird that I want one? 🤣

• Neural Dynamics: Neuralink is not the brain-to-computer interface company installing nanodevices directly into people’s brains. None of that weak-ass ‘wearables’ stuff here! Lol.

Conclusion: So what should you do?

Hope you enjoyed this whirlwind, very high-level (for now) tour of the bleeding edge as much as I’ve enjoyed the research. In future posts, I’ll continue to do (shorter) round-ups and ‘market maps’ like this especially as new players/dynamics emerge, and I’ll also do more focused posts either on one category, one use case, or even just one company. I hope to get speculative non-fiction and sci-fi contributions from prominent voices in the community as well. If you know someone I should profile, or if you’d like to contribute content here, please email me at [email protected] or DM me on X or LinkedIn.

After that… if you’re not already, you should build something! What this research has shown me more than anything is that what seems like the far future is actually very close and approaching faster than we realize. Meanwhile, the deep tech opportunity space is still massive with decades ahead of it, and technical barriers to entry are dropping quickly.

Hey… maybe your crazy idea isn’t that crazy after all?