Philip Johnston is the CEO of Starcloud. He put an Nvidia H100 chip in space and gave a TED talk about it. As you do if you're responsible for building the infrastructure for space-based data centers. 

Elon Musk was not the first. He follows in the footsteps of Mr Johnston. And so, rather than Mr SpaceX, our first technology reaction video is this TED talk from San Francisco.   

We watched it for the first time. Live. On TV. 

This is not theoretical. It's also not up to date. Philip filmed this in October 2025. Starcloud have already launched the Nvidia H100 on a Falcon 9 up into space. 

It's happening disruptors and curious minds. It's happening. 

Philip predicts most data centers will be in space within 10 years. 

We agree. 

Please enjoy the show.

Cheers, 

Mark & Jeremy.

--

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

• Email: [email protected]

Philip Metzger spent nearly 30 years at NASA. His conclusion is simple and it is hard to shake: if we do not democratise the ownership of space before the economy explodes, a handful of people will control a resource worth billions of times the current global economy.

Democracy does not survive that.

This episode covers:

• Every rocket landing on the moon blows dust at six times the speed of a bullet. No country has yet agreed on what to do about it. After enough launches, antennas jam, sensors fail, and hardware designed for decades dies in years
• Philip's calculations show it will be cheaper to build large AI data centres in space than on Earth within ten years. Jeff Bezos, Eric Schmidt, and Elon Musk have all independently arrived at the same number
• If launch rates reach ten Starships per day, we may hit the stratospheric damage limit. SpaceX is planning one launch every two minutes globally. There is no international framework for this yet
• Fully autonomous luxury communism: what it actually means, why space resources make it possible, and why it requires democratising ownership right now
• Every economic revolution has produced slavery, exploitation, and war. Philip does not think we have matured enough to avoid that this time
• Why a student robotics competition in Mexico might be one of the most important space programmes on the planet

Note: Philip's laptop overheated mid-conversation. Part two covers the Fermi paradox, Gerald O'Neill, and what it means to be human when civilisation is space-based.

--

Take your Thinking Further. 

Stephen Hawking Center: https://sciences.ucf.edu/physics/microgravity/lab/

Philip X: https://x.com/drphiltill

-

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

Email: [email protected]

--

TIMESTAMPS

(00:00) Introduction to Space Exploration and Economics

(01:26) NASA's Role in Future Space Exploration

(06:45) Impact of Rocket Exhaust on Lunar Soil

(14:39) Geopolitical Challenges in Space

(23:39) Democratizing Space for Future Generations

(33:45) Emergent Forces vs. Hierarchical Forces

(34:08) Exploring Microgravity Applications

(38:39) Rapid Fire Space Technology Opinions

(44:02) The Future of Humans and Technology

Iona's drone takes off vertically, tilts its propellers, and flies like a fixed-wing aircraft. 126km/h. 100km range. 20kg payload. Cargo bay opens from the bottom on arrival. Nobody needs to be at the other end to receive it. At scale, under $1 per parcel.

Etienne Louvet built the regulatory compliance layer directly into the flight software. The drone cannot take off if the criteria are not met. That is how you demonstrate to an aviation authority that you are serious.

And it is why Iona deliberately operates under the strictest regulators in the world rather than going somewhere easy, if you can self-authorize under EASA SORA and the UK CAA, you are scalable anywhere. Everything else is a marketing stunt.

This episode covers:

• Why drone delivery has been promised since 2013 and is only viable now: the EASA SORA regulatory framework (2021) gave companies the certainty to know what they needed to build
• Ireland, UK, and the US Part 108 framework compared, and why regulatory iteration speed is the variable that determines which companies survive
• BVLOS and self-authorization explained: what beyond visual line of sight actually unlocks and how Iona's software makes compliance non-negotiable at the hardware level
• Why remote communities do not resist drones: they want access, not isolation, and the density of rural areas makes the dystopian sky-full-of-drones scenario mathematically impossible
• No cameras, no personal data: why Iona collects only aeronautical and robotics data and why that is the right call
• The operating system vision: drones are the starting point, ground robots are next, and the software stack is vehicle-agnostic

--

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

• Email: [email protected]

--

TIMESTAMPS

(00:00) Intro 

(01:50) How much weight can drones carry

(02:29) What counts as light cargo 

(06:51) How drone regulations actually work 

(13:04) Self-assessment and risk management 

(14:12) Getting municipalities to say yes 

(16:38) Weather problems 

(19:48) Where Iona Drones is now 

(20:58) Maximizing payload capacity 

(21:58) Drone design choices 

(23:27) BVLOS explained 

(26:08) Drones and privacy concerns 

(30:45) Implementing drones in existing logistics 

(35:02) Where autonomous delivery is headed 

(39:30) Technology and human progress

Can you use AI to think better or think more critically? Philosopher Pia Lauritzen says no. The second we give up to the shortcut use AI, we are letting go of the very basic condition that forces us to think.

When we ask if machines can think, the first question should be: why do humans think? Why do we think?

For Pia, it is fairly simple. We think because we know there is something we do not know. We have a problem. There is a gap. A gap between what I know and what I want to know. 

So I have to start thinking. That is why I ask these questions and that is why I put up with this pain in my head of trying to figure something out that I do not know.

The machine does not have that problem. It does not know that it does not know. 

It is like an animal. 

It does not know that it does not know. 

Of course it is a matter of how you understand thinking. But if you consult the old thinkers and not just the engineers and technologists, then you will have a really hard time finding anyone who would say that a machine could ever think. 

And if it cannot think itself, why should it be able to help us think? We are the only ones who know how to do that.

This is the core problem. AI feels helpful. It removes the discomfort of not knowing where to start. It fills the blank sheet. But that discomfort is not a bug. 

That discomfort is the feature. That discomfort is what thinking is.

And it is at this point that I am reminded of the scene in Con Air. Define irony.

Please enjoy the show.

Cheers, 

Mark & Jeremy.

PS: Subscribe so other curious minds like you can find our channel.

--

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

Email: [email protected]

We taught technology to generate music before we taught it how to assign fair credit to musicians. Nicholas Ponari explains why Overtune rejects one-prompt AI music generation in favor of human-in-the-loop creation. 

Unlike platforms trained on scraped catalogs, Overtune’s AI is built on licensed music, starting with ~20,000 loops produced in-house. Producers can submit stems voluntarily, creating a clean foundation for ethical training and attribution.

The platform uses vector-based audio embeddings to measure how much each stem contributes to a generated track. This enables automated attribution and proportional royalty distribution when songs are commercialized. 

Contributions are weighted mathematically, with clear thresholds to credit primary and secondary influences while avoiding excessive fragmentation 

Please enjoy the show.

Cheers,

Mark and Jeremy

PS: Subscribe so other curious minds like you can find our channel.

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

• Email: [email protected]

Marketing funnels don't exist. They never did. The internet just convinced us they were real. Meta, Google, OpenAI and a supporting cast of billionaire sociopaths figured out they could control distribution and black-box your customers.

Hurrah. Humanity forgot to read the small print. 

Now you're running a business where you don't even know who your customer is.

Well here’s the AI-shaped healthcheck: Your customer is OpenAI.

You're paying 3-15% for a digital presence you don't need. It's called the Silicon Valley tax. You're burning money to keep VCs rich while platforms add another layer of black box between you and the people you serve.

The alternative? Network methodology. Someone you know, or someone who knows someone you know. That's it.

Funnels were invented to sell marketing. Networks are how humans actually work. 

We've been doing it since we had prefrontal cortexes.

Everything that's real is analog. 

That's true for business too.

Welcome to the marketing jungle. 

The year is 2026, and if you don’t know who the sucker at the table is… you probably shouldn’t be playing the stakes. 

Please enjoy the reality check.

Cheers,

Mark and Jeremy. PS: Keep thinking on paper. They don’t want you to, that’s why you must. 

--

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

• Email: [email protected]

--

Matt Kinsella runs Infleqtion, a company building quantum computers. The biggest misconception about quantum computing is that it will replace classical computing. It won't. 

Quantum processors will sit above GPUs in data centers the same way GPUs sit above CPUs today. 

NVIDIA just built the bridge to make this work. It's called NVQ Link, and it changes how we think about the future of compute.

NVIDIA announced NVQ Link in October 2024. It's the bridge between quantum computers and classical GPU clusters. Workloads pass seamlessly between them.

Here's how it works in practice. Infleqtion and NVIDIA solved something called the Anderson Impurity Model - a photovoltaic problem in material science. Parts of it were solved on a GPU cluster. Parts that couldn't be solved by GPUs were solved on Infleqtion's quantum computer. 

Then they recombined to give the answer. 

This isn't commercially useful yet. But expand that over time and you could be looking at the future data center. One with three layers. CPUs at the bottom for general computing. GPUs in the middle for parallel processing and AI. QPUs at the top for problems that are quantum mechanical in nature. 

Workloads come in, get chopped up, each piece goes to the part of the stack best suited to solve it. Then results recombine.

This is already happening. Infleqtion just announced a contract with the Army called Sapient Secure AI for PNT - position, navigation, and timing. It runs their quantum-inspired software on NVIDIA's Jetson edge GPUs. 

Small GPUs that don't have much memory. 

The software lets them ingest far more streaming data than normal. Video, speed, inertial motion. Then it recreates what GPS gives you - where you are in the world - by extrapolating from all those signals. Without GPS.

Please enjoy the show.

Cheers, 

Mark & Jeremy.

--

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

• Email: [email protected]

--

Before anyone was asking whether commercial space stations could turn a profit, three ex-NASA engineers were buying satellite components on Amazon and launching them into orbit.

In this episode, Mark and Jeremy learn about Planet Labs — the company that built a constellation of 150 tiny satellites out of laptop batteries and off-the-shelf phone parts, and ended up creating what the authors call a real-time accounting system of the Earth.

Their satellites were the first to spot Russian troops massing on the Ukrainian border before the invasion.

The business model? Selling that data to governments. Which is how the internet started too.

Then they get into the real question nobody wants to answer: what is the killer app for space? Hotels in orbit sound exciting until you read that a 2017 study found commercial space stations were unlikely to be economically viable by 2025.

Finally, they learn how the Le Chatelier principle says the short run is a poor guide to the long run.

The space economy may look underwhelming today. That might not mean anything about what comes next.

We're reading Space to Grow by Matthew Weinzierl and Brendan Rosseau. Come think on paper with us.

--

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

• Email: [email protected]

--

TIMESTAMPS

(00:00⁠) Trailer 

⁠(01:35⁠) No Dust Jackets ⁠

(02:00⁠) Name Jeremy's Astronaut ⁠

(03:52⁠) What Is The Product Market Fit For Space? ⁠

(05:26⁠) Satellites And The Le Chatelier Principle ⁠

(09:00⁠) Planet's Dove Satellites ⁠

(16:38⁠) Satellites For Climate ⁠

(18:28⁠) John Lewis ⁠

(22:30⁠) Ronald Reagan & Carl Sagan ⁠

(26:42⁠) Inflatable ISS Modules

Two thirds of data centre outages are caused by someone pressing the wrong switch. Not a cyberattack. Not a hardware failure. A person, in a building full of thousands of switches, turning off the wrong one.

Shapol and his co-founder built four reusable rocket missions before they turned to data centres. Their conclusion: the most critical infrastructure on the planet is still being operated with Word documents, printed manuals, and procedures written by hand weeks in advance - documents that are rarely updated as hardware changes, and that rely entirely on humans not making mistakes.

Entangl ingests every design document, every circuit, every asset, and produces step-by-step operating instructions deterministically. Same input, same output, every time — like an aircraft autopilot. No variability. No hallucinations. Because in infrastructure AI, hallucination is not a feature. It is a catastrophic failure.

This episode covers:

• Why 18-month generator lead times force data centres to operate without backup power, and what that means for the next outage
• How Entangl's system detects every asset in a data centre, generates every possible operating procedure automatically, and updates them in real time as hardware changes
• Deterministic vs generative AI: why ChatGPT-style creativity is exactly what you do not want in a system that controls whether hospitals stay online
• The Dunning-Kruger problem in data centre operations: why experienced engineers are often the most dangerous ones
• VR training: how engineers learn to work on live systems without touching them
• The Challenger framing: do humans always have the final say, and should they?
• Space-based data centres: Shapol is in the same YC cohort as Star Cloud and refers them directly. His view on when heavy industry moves off Earth

--

Other ways to connect with us:

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

Email: [email protected]

--

TIMESTAMPS

(00:00) Trailer

(02:17) From rocket launches to data center automation

(06:00) How Entangl integrates with building monitoring systems

(08:34) Data Center Design constraints: How AI fixes it

(15:37) AI, Dunning Kruger And Hallucinations

(21:42) Will humans always have the final say in data centers?

(24:53) Space-based data centers and solar power

(25:04) Kevin Kelly's question: What should humans become?

SpaceX launches 135 rockets a year. NASA's shuttles launched five. SpaceX delivers cargo to orbit for $2,800 per kilo. The shuttles cost $90,000. In fifteen years, one company did what a government agency couldn't do in sixty.

Mark and Jeremy work through chapters one to three of Space to Grow by Matthew Weinzierl and Brendan Rosseau, covering the three-act history of NASA and the birth of the private space industry.

This episode covers:

• How the Apollo programme's end and the Columbia disaster forced NASA to open the gates to private companies
• The COTS contracts that shifted financial risk onto private companies and drove innovation
• Elon Musk's failed Russia trip and the decision to build SpaceX from scratch
• Three rocket explosions, $100 million left, and a fourth rocket built from spare parts in a shed
• Why someone had to climb inside a rocket mid-flight to hammer out the dents
• SpaceX suing the Air Force over launch contracts and winning · Jeff Bezos at five years old watching Apollo, then quietly building Blue Origin for a decade
• The four principles behind SpaceX's success: iteration, vertical integration, reusability, and culture
• "I'm going to build the Honda Civic of space rockets"

--

• ⁠Listen to every podcast⁠

• Follow us on ⁠Instagram⁠

• Follow us on ⁠X⁠

• Follow Mark on ⁠LinkedIn⁠

• Follow Jeremy on ⁠LinkedIn⁠

• Read our ⁠Substack⁠

• Email: [email protected]--

--

TIMESTAMPS

(00:00) Trailer

(01:02) Space To Grow

(01:55) Incorporate Space Into Your Thinking

(03:28) The Apollo Program Ends

(05:43) The NASA Budget & Shuttle Launches

(07:51) Bush & The Aldridge Commission

(08:36) COTS (Commercial Orbital Transportation Services)

(10:27) Blue Origin, Bezos & O'Neill

(14:40) A Quick History Of SpaceX

(18:23) Falcon Blows Up

(20:24) Elon Sues The Airforce

(22:04) SpaceX Launch Costs

(23:45) The Honda Civic Of Space Rockets