Thinking On Paper

AI Inference, Wasted GPUs And The Cost Of Every Token

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 [https://thinkingonpaperpodcast.substack.com/] 🫵 C⁠hoose your own technology adventure ⁠ 📺  ⁠Watch our beautiful faces on YouTube ⁠ [https://www.youtube.com/@thinkingonpaper/videos] 🎧 R⁠emember Steve Jobs on APPLE⁠ [https://podcasts.apple.com/us/podcast/thinking-on-paper/id1713227258] 📺 ⁠Get clips and exclusive videos on Instagram ⁠ [https://www.instagram.com/toptechpodcast/%20%C2%A0] -- 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

The Tasty Truth Of Cultivated Meat - Bruce Friedrich

Could cultivated meat replace industrial farming without asking billions of people to change what they eat? And save all the cows on the way? Bruce Friedrich, founder of the Good Food Institute and author of Meat, explains why the future of food may depend less on changing human behaviour and more on changing how meat is produced. For more than 12,000 years, humans have relied on animal agriculture to produce meat. Bruce argues that this system is reaching its limits. We explore how cultivated meat is made, why it is attracting support from governments and major food companies, and why some of the biggest meat eaters are also the most open to the technology. Along the way, we discuss the environmental cost of livestock, antibiotic resistance, food security, infrastructure, consumer psychology, regulation, and whether this technology could become as transformative as renewable energy or electric vehicles. We also examine what happens to farmers, why culture may be harder to change than technology, and whether cultivated meat can ever reach price parity with conventional meat. If the next agricultural revolution is already underway, it may not begin on a farm. 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 [https://thinkingonpaperpodcast.substack.com/] 🫵 Choose your own technology adventure [www.thinkingonpaper.xyz] 📺  Watch our beautiful faces on YouTube [https://www.youtube.com/@thinkingonpaper/videos] 🎧 Remember Steve Jobs on APPLE [https://podcasts.apple.com/us/podcast/thinking-on-paper/id1713227258] 📺 Get clips and exclusive videos on Instagram [https://www.instagram.com/toptechpodcast/  ] -- Chapters (00:00) Who Eats Cultivated Meat? (02:37) Cultural Significance of Meat (06:09) What Is Cultivated Meat? (08:25) The Science Behind Cultivated Meat (10:48) How Much Does Cultivated Meat Cost? (14:46) What does cultivated meat taste like? (17:10) Emerging Companies in the Lab Meat Industry (19:08) Industry Support for Alternative Meats (21:06) Government Role in Innovation (24:40) The Future of Animal Agriculture (28:25) Agricultural Policies for Sustainability (31:41) What Are Fermented Meats? (35:48) Meat And Antibiotic Resistance (39:32) The Invisible Costs of Meat Production (42:26) The Future of Cultivated Meat

Asteroid Mining Is Easier Than You Think - Matthew Gialich, Astroforge

Asteroid mining sounds insane until you speak to AstroForge CEO Matthew Gialich. Then it makes perfect sense. Matthew’s team at AstroForge builds spacecraft to mine metallic M-type asteroids for platinum group metals, the unglamorous but essential metals inside phones, cars, chips, electronics and much more. AstroForge is one of the few companies trying to make space mining real, targeting metal-rich asteroids that could contain platinum, palladium, iridium and other PGMs. Why? Earth’s resources are getting harder, deeper and more expensive to reach. The good stuff is not sitting neatly on the surface. A lot of it is buried, depleted, regulated or uneconomic. In this episode, Matthew explains why asteroid mining may now be technically and economically possible, why Planetary Resources may have been too early, how SpaceX changed the capital story for space startups, and why AstroForge’s first deep-space spacecraft failed after travelling nearly a million miles from Earth. 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 [https://thinkingonpaperpodcast.substack.com/⁠] 🫵 Choose your own technology adventure [www.thinkingonpaper.xyz] 📺  Watch our beautiful faces on YouTube [https://www.youtube.com/@thinkingonpaper/videos] 🎧 Remember Steve Jobs on APPLE [https://podcasts.apple.com/us/podcast/thinking-on-paper/id1713227258] 📺 Get clips and exclusive videos on Instagram [https://www.instagram.com/toptechpodcast/⁠ ⁠ ] -- Chapters (00:00) Asteroid Mining Trailer (02:51) The Economic Necessity of Asteroid Mining (08:37) Lessons from Planetary Resources (10:43) Risk and Innovation (15:26) Deep Space Two (19:05) The Quest for Asteroid Exploration (22:20) Aliens & Life Beyond Earth (24:17) Ownership and Ethics in Space Mining (26:21) The Next Challenges in Asteroid Mining (27:56) Changing Earth's Economy (30:13) The Role of Capital (32:32) Matt's Vision for Space Exploration (35:06) The Drive to Explore the Universe (36:55) NASA's Evolution (39:17) Inside Astroforge (42:53) The Complexity of Space Engineering (44:25) Data Centers in Space (47:30) The Limitations of AI in Space Engineering

How To Build Your Own Quantum Computing Hype Filter

Quantum computers could change medicine, material science and AI. Unfortunately, it's impossible to know how because the industry is buried under hype. In today’s show, Dr. Bob Sutor takes us on a masterclass through the hype and conflicting headlines to the reality of quantum technology today.  By the end of the show you’ll know what’s fact and fiction, where the industry is heading and how to protect yourself from AI slop masquerading as insight. We also learn about: * Why “quantum supremacy” and “quantum advantage” claims should be treated carefully * Why quantum computing is still in its prehistory * How engineering discipline is changing the field * What IBM and Cleveland Clinic’s protein simulation work shows about quantum chemistry * Why money, sovereignty and technology are driving the quantum industry * How governments are funding quantum computing in the United States, France, the UK, Finland and elsewhere * What China may be doing in quantum computing * Which industries are doing serious quantum research * Why there are many quantum hardware approaches and no clear winner yet * Whether helium-3 supply matters for quantum computing Bob also explains why quantum computers will be useful for specific classes of difficult problems where classical computers struggle, especially once systems become larger, more reliable and fault tolerant. The conversation ends with practical advice on separating the quantum noise from the Thinking On Paper signal. 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 [https://thinkingonpaperpodcast.substack.com/] 🫵 Choose your own technology adventure [www.thinkingonpaper.xyz] 📺  Watch our beautiful faces on YouTube [https://www.youtube.com/@thinkingonpaper/videos] 🎧 Remember Steve Jobs on APPLE: [https://podcasts.apple.com/us/podcast/thinking-on-paper/id1713227258] 📺 Get clips and exclusive videos on Instagram [https://www.instagram.com/toptechpodcast] (00:00) AI Quantum Slop (00:40) Welcome To The Show (06:04) When Quantum Computers Finally Become Useful (10:13) Why Governments Are Throwing Money at Quantum (15:53) Is China Ahead in Quantum? (18:48) Where Quantum Might Actually Matter (27:11) Can Quantum Help Fix Climate Change? (28:35) Why Battery Companies Care About Quantum (30:31) Why Quantum Doesn’t Belong in the IT Department (31:53) Who’s Doing the Real Work in Quantum? (32:40) Why Quantum Companies Need Real Customers (38:19) The Funding Problem Behind Quantum Progress (45:22) Does Quantum Computing Need Helium-3? (48:01) Will We Need a Quantum Computer Matchmaker? (50:38) How to Spot Quantum Hype Before You Share It

AI Search Will Destroy Google

Have you used Google Search recently? Exactly. Most companies, and most people, still think about Google when they think about search. They’re still spending heavily to rank there and paying for the ads around it. But more people are asking ChatGPT, Claude and Gemini what to buy, read, use or trust. SEO isn’t disappearing. It’s evolving into GEO. Awad Sayeed, co-founder and CTO of Parsnipp AI, joins Thinking on Paper to explain generative engine optimisation, or GEO, and how companies can become more visible inside ChatGPT, Claude, Gemini and other AI answer engines. Traditional SEO focuses on keywords, backlinks and rankings. GEO is more dependent on context: who the user is, what they’ve already asked, what they’re trying to achieve and how an AI system retrieves and combines information. In this episode, we discuss: * How generative engine optimisation differs from SEO * Why context matters more than keywords in AI search * How ChatGPT, Claude and Gemini use information differently * What persona-based agents reveal about brand visibility * How structured data helps AI systems understand websites * Why comparison pages and clear product information matter * What black-hat GEO could look like * How AI-generated content could pollute the internet * Whether brands should create separate experiences for humans and AI agents * How advertising may develop inside AI assistants Awad argues that GEO doesn’t replace SEO. Strong websites, useful content and clear structure still matter. But companies now need to think about whether AI systems can retrieve, interpret and recommend their information in the right context. And as this is Thinking On Paper, we ask about the human impact, the wider change in the structure of the internet, trust, data and consumerism.  Please enjoy the show. -- 🏠 Buy us a beer on Substack [⁠⁠https://thinkingonpaperpodcast.substack.com/⁠] 🫵 Choose your own technology adventure [www.thinkingonpaper.xyz] 📺  Watch our beautiful faces on YouTube [https://www.youtube.com/@thinkingonpaper/videos] 🎧 Remember Steve Jobs on APPLE [https://podcasts.apple.com/us/podcast/thinking-on-paper/id1713227258] 📺 Get clips and exclusive videos on Instagram [https://www.instagram.com/toptechpodcast/⁠ ⁠ ] -- Chapters (00:00) Introduction to Generative Engine Optimization (03:36) Understanding Persona-Based Agents (06:23) The Transition from SEO to GEO (09:06) Context in LLMs and GEO (11:41) Black Hat Strategies in GEO (14:22) The Future of the Internet (16:58) Advertising in the Age of GEO (19:37) The Impact of GEO (28:22) The Evolution of AI Models (29:03) Integrating AI into Business Strategies (29:52) Agents vs. Humans (32:10) The Future of SEO and GEO (34:08) Tools for Visibility and Analytics in AI (36:00) Customer-Driven Development (39:23) The Role of Storytelling in GEO (42:04) Model Transparency and the Future of AI