Thinking On Paper

We read the National Space Transportation Policy in full. Signed 20 August 2026, it shreds the 2013 directive and rewrites how the US gets into space, moves around up there, and comes back down.

The War Department is all over it. Launch ranges, reentry infrastructure, a clause about accommodating launches within 48 hours of need. You don't build a 48-hour capability to do science. Somebody expects to lose a satellite and need a replacement up fast.

Here's what we think stops it. Not rockets. Not pads. Demand. Most of what goes up is Starlink, everyone else crams onto rideshares, and a target of 1,000 assumes five times more reason to be up there than anyone currently has. The document is a sales pitch with the government as a customer.

We also get into the Mars carve-out and what it says about who America needs, the ISS retiring the same year and China’s plans to land on the moon.

--
Thinking on Paper is a technology podcast about AI, Space, quantum computing, science, and the systems shaping your life. 
🏠 ⁠Buy us a beer on Substack⁠
🫵 C⁠hoose your own technology adventure ⁠
📺  ⁠Watch our beautiful faces on YouTube ⁠
🎧 R⁠emember Steve Jobs on APPLE⁠
📺 ⁠Get clips and exclusive videos on Instagram ⁠
--
Chapters

(00:00) Introduction to the US Space Transportation Policy
(03:20) US Launches in 2025 and 2026
(04:45) Private Industry and Market Demand
(06:24) International Competition and Collaboration
(07:55) Infrastructure Needs for Increased Launches
(10:00) Government Coordination and Regulatory Frameworks
(12:02) Market Opportunities and Ride-Sharing in Space
(15:13) Future Missions to Mars and Lunar Surface
(18:43) International Efforts and the Race to Mars
(20:48) Implications of the 2030 Space Goals
(23:22) Summary and Next Steps

Why do we stop asking questions as adults? Philosopher Pia Lauritzen has analysed 30,000 of them, and says nothing was ever lost.

The ability stays. What changes is the price. Every question admits you don't know something, and nobody gets promoted for ignorance. Ask in the wrong room and curiosity starts to feel like an interrogation, which is why most of us learn to keep the question to ourselves.

She proves it in keynotes. One minute, two rules, and every person in the audience writes one. They could do it the whole time.

Full episode with Pia Lauritzen on questions, curiosity and whether AI can think at www.thinkingonpaper.xyz

Be curious, stay disruptive.

--
Thinking on Paper is a technology podcast about AI, Space, quantum computing, science, and the systems shaping your life. 

🏠 ⁠Buy us a beer on Substack⁠
🫵 C⁠hoose your own technology adventure ⁠
📺  ⁠Watch our beautiful faces on YouTube ⁠
🎧 R⁠emember Steve Jobs on APPLE⁠
📺 ⁠Get clips and exclusive videos on Instagram ⁠

Nicholas Ponari says AI music makes live performance more valuable, not less. Mark Fielding calls the same future a hellscape.

Ponari's scenario: by 2045 his son types a script into a generator, gets a feature film in the style of Kubrick crossed with Tarantino, and hits publish. 

Everyone becomes a creator in every medium. Nobody can consume a fraction of it, so AI becomes the largest curator in history.

Mark's objection is that shared culture dies with it. No song anyone else has heard. No film anyone else has seen. Nothing left to talk about.

Ponari's answer is scarcity. When everything is generated, the things that cannot be, a room, a crowd, live music, become more valuable than ever.

Who's right?

Full episode at thinkingonpaper.xyz.
Be curious, stay disruptive.

--
Thinking on Paper is a technology podcast about AI, Space, quantum computing, science, and the systems shaping your life.

Thinking on Paper is a technology podcast about AI, Space, quantum computing, science, and the systems shaping your life. 
🏠 ⁠Buy us a beer on Substack⁠
🫵 C⁠hoose your own technology adventure ⁠
📺  ⁠Watch our beautiful faces on YouTube ⁠
🎧 R⁠emember Steve Jobs on APPLE⁠
📺 ⁠Get clips and exclusive videos on Instagram ⁠

The US and China are racing to build AI systems that could lead to superintelligence before any enforceable global rules exist to inspect them, limit them or verify what they can do.

Mark and Jeremy Think On Paper about The Hard Question of AI, an essay by Planet CEO Will Marshall. Marshall argues that AI governance now needs the same kind of verification, trust and global protection as nuclear weapons.

That means more than voluntary safety tests and China and America laying down their AGI swords. And since that is unlikely, it could mean monitoring chips, tracking data-centre construction and energy use, inspecting frontier AI labs, protecting whistleblowers and agreeing international red lines around recursive self-improvement.

But the nuclear comparison has many challenges. Weapons-grade material is scarce and difficult to hide. Just ask Iran. AI software can spread quickly and cross borders. Any Tom, Dick or Harry could end up being the first to create superintelligence. 

At the centre of the episode is one question: what would AI’s Cuban Missile Crisis look like? What event would force the US and China to cooperate? Really cooperate. Lip service won’t calm existential risk. 

We also discuss Sam Altman, Dario Amodei and Demis Hassabis, the incentives behind AI safety warnings, the limits of voluntary regulation, the RAND verification framework, recursive self-improvement, the Fermi paradox and whether governments can build the institutions required to manage AI superintelligence.

Hint: they can’t.

Please enjoy the show. And don’t have nightmares.

--

Thinking on Paper is a technology podcast about AI, Space, quantum computing, science, and the systems shaping your life. 
🏠 ⁠Buy us a beer on Substack⁠
🫵 C⁠hoose your own technology adventure ⁠
📺  ⁠Watch our beautiful faces on YouTube ⁠
🎧 R⁠emember Steve Jobs on APPLE⁠
📺 ⁠Get clips and exclusive videos on Instagram ⁠

--

Chapters

00:00 Trailer 
01:38 AI And Existential Risk
02:40 OpenAI, Anthropic & Google
04:40 Nuclear meltdown or sharks?
05:50 Why is everyone speaking about the end of humanity
08:00 RSI - Closed loop recursive self improvement
10:00 A ghostbuster copyright interlude
19:39 Why America And China Are In An AI Race
23:51 The AI Cuban Missile Crisis
28:07 The AI Supply Chain
35:09 USA And China And The AI DeadEnd
40:36 The Nuclear Framework

AI companies are splashing billions on more GPUs before they use the ones they already have. Many sit idle, others use a fraction of their potential. 

For the hyperscalers, cloud providers and AI companies, buying more GPUs does not solve much if the GPUs already in the rack spend too much time waiting for the rest of the system.

The next battle in the AI infrastructure wars is efficiency: GPU utilization, NICs, memory movement, networking, CPUs and software that keeps expensive hardware doing useful work.
Moshe Tanach runs NeuReality AI, a company building infrastructure for AI inference. In this conversation, he explains why the next wave of AI infrastructure depends on GPU utilization, NICs, memory movement, networking and software efficiency, not just buying more chips.

This is a technical conversation about AI inference, infrastructure efficiency and the hardware layer beneath ChatGPT, Claude, Gemini and the AI tools now being built into business software.

Please enjoy the show.

Thinking on Paper is a technology podcast about AI, Space, quantum computing, science, and the systems shaping your life. 
🏠 ⁠Buy us a beer on Substack⁠
🫵 C⁠hoose your own technology adventure ⁠
📺  ⁠Watch our beautiful faces on YouTube ⁠
🎧 R⁠emember Steve Jobs on APPLE⁠
📺 ⁠Get clips and exclusive videos on Instagram ⁠

--
Chapters

(00:00) The GPU challenge
(01:27) Training Vs Inference
(05:45) Memory & Keeping The GPU Busy
(07:31) Deep Seek, Blackwell & Ruben
(09:03) How Ripe Is CPU For Reinvention?
(11:03) AI Agents Run On CPUs
(12:51) Do We Need All The Computation?
(14:35) GPU Utilization Rates
(16:50) Tokens In A Context Window
(20:59) Data, Knowledge & Wisdom
(23:56) Why Are Your GPUs Busy? (A Truck Analogy)
(29:07) Hyperscalers
(31:59) The Decode Phase
(36:47) What More Efficient GPUs Means For The User