Konstantin Saifoulline

David Patterson helped invent RISC and RAID, won the Turing Award, and worked on Google's TPU. He explains why power is now AI's hardest limit, what replaces Moore's Law, and who could win the AI chip race.

CHAPTERS
00:00 Power Is the Problem
03:10 The Power vs Performance Wall
05:46 The Memory Breakthrough AI Needs
09:07 Moore's Law Is Slowing. Now What?
11:59 How RISC Changed Every Chip You Use
14:54 Inside Google's TPU Bet
18:02 What AI Chips Look Like Next
21:01 How Do You Cool Chips This Hot?
24:06 Why AI Is Moving to Lower Precision
26:58 "The Appetite for AI Is Ravenous"
29:47 Who Wins the AI Hardware Race?

China has robots running its factories with fewer and fewer workers. American manufacturers can’t even afford their own equipment.

Matt Parlmer left software engineering during COVID to found General Fabrications (GenFab), and he says the whole system is broken: “Everything is too expensive.”

Chapters
00:00 From Software to Manufacturing
03:21 Why Manufacturing Is Broken
08:25 GenFab’s R&D: “Factorio IRL”
14:04 What Software Engineers Get Wrong About Hardware
17:20 Why CAD Will Go Open Source
22:23 The Capital Problem Killing US Manufacturers
26:29 The Next Decade: Fully Instrumented Factories
29:20 How China Changed Manufacturing Forever
32:07 The Future of American Factory Jobs
34:48 AI, Physics Models & Their Limits
39:57 Are the Big AI Labs Overvalued?
47:58 Elon Musk’s Terafab & Cheaper GPUs
53:01 Why Robots Aren’t in American Factories Yet

#Manufacturing #Robotics #AI #China #Hardware

Pearse Coyle was founding CEO, first investor or first head of sales in 11 tech ventures, and now backs deep tech at seed. His one rule for hardware founders: don't die. Here's how to stretch cash, use suppliers and grants, and pick the right first investor.

CHAPTERS
00:00 "Don't Die": The Only Rule
03:02 Why Hardware Eats Cash
05:48 How to Spend Almost Nothing Early
09:10 Find Customers Before You Build
12:11 Suppliers Can Make or Break You
15:09 The Funding Walls Every Startup Hits
17:57 Why Early Adopters Matter
20:47 How to De-Risk a Venture
23:58 Using Suppliers to Help Cash Flow
26:48 Climbing the Funding Walls
29:50 The Grant Money Most Founders Miss
32:07 Why Early-Stage Funding Is Hard
35:27 Why Realistic Forecasts Win Investors
39:30 How to Spot a Good Investor
43:29 Why Your First Check Matters Most
47:27 Managing Your Investors
51:45 Advice for Founders

#Startups #HardwareStartups #VentureCapital

Tom Dietterich has worked in machine learning since the very first ML workshop in 1980. He explains why today's AI is brilliant until it meets something new, why we still can't guarantee its outputs, and who should be watching it.

CHAPTERS
00:00 Where "Machine Learning" Came From
10:06 Neural Networks vs Kernel Methods: The Rise and Fall
20:57 Where AI Research Is Heading Next
28:51 What We Still Don't Understand About Deep Learning
30:12 Why AI Safety Is a Moving Target
32:59 Why Humans Still Need to Watch AI
35:02 AI's Biggest Weakness: Anything New
39:30 Why AI Forgets What It Learned
40:46 When AI Fails, Who's to Blame?
43:17 Who Monitors Superhuman AI?
49:20 AI Agents, Regulation and What Comes Next

#AI #MachineLearning #AISafety

AI is moving beyond screens and into the physical world.

From autonomous robots to AI-powered customer support, government automation, and building habitats in space, the next wave of AI could transform much more than software.

I sat down with Ben Holfeld to discuss how technology is changing the way we build businesses, distribute products, automate industries, and explore space.

We cover why distribution is becoming increasingly important, the rise of physical AI, real-world applications of automation, and how robots could eventually help humanity expand beyond Earth.

Chapters
00:00 The Evolution of App Distribution
05:56 The Rise of Physical AI
11:59 AI in Customer Support and Enterprise Solutions
18:11 The Future of Space Exploration and Robotics

Host: Konstantin Saifoulline
Guest: Ben Holfeld

https://konstantinseifoulline.com
https://airsup.ai

Subscribe for conversations about AI, robotics, space exploration, engineering, and the technologies shaping our future.

#AI #Robotics #PhysicalAI #Automation #SpaceExploration

AI companies are racing to build bigger data centers. But the biggest bottleneck might not be chips. It's electricity.

I sat down with Nat Bullard, co-founder and Chief Strategy Officer of Halcyon and former Chief Content Officer at BloombergNEF, to discuss the physical limits of scaling AI.

00:00 Why AI can't scale without electricity
02:30 Ranking AI's biggest infrastructure bottlenecks
04:02 Why we're already hitting power limits
05:07 Building power infrastructure outside the grid
07:54 How xAI and Meta are securing electricity
10:16 OpenAI, Anthropic and the race for power
12:33 Backup power vs bridge power vs primary power
13:55 The next five years of AI energy economics
16:22 500 proposed gas plants and the future energy mix
18:17 Why nuclear power may take longer to scale
20:54 Gas vs nuclear vs renewables for the long term
23:47 Building data centers in 12 months instead of 36
25:01 What happens when GPUs become obsolete?
27:23 Reusing waste heat from AI data centers
30:59 Could data centers become part of future cities?
33:54 How many announced data centers will actually get built?
36:44 Are data centers in space realistic?
38:32 US vs China: Who can scale electricity faster?
40:42 Final thoughts

What happens when billions of AI agents start doing real work?

Nick Vasilescu is the co-founder of Orgo, which is building cloud computers for AI agents. His thesis is simple: every AI worker will eventually need its own persistent computer, just like every human worker does today.

Nick Vasilescu: @nickvasiles
https://www.orgo.ai

00:00 Why 10 billion AI workers will need computers
03:18 Orgo vs OpenAI and Grokbot
06:02 Why agents need real computers
08:07 How Orgo went viral
11:48 Why viral content stopped converting
13:57 The best AI agent tools right now
17:42 Can smaller AI companies survive OpenAI?
19:33 What San Francisco is building for the next 5–10 years
20:50 The business opportunity in AI workers
21:19 Generative UI and the end of static interfaces
21:48 Voice becomes the interface for AI
23:53 Is “Her” becoming real?
25:53 What comes after phones and laptops?
27:09 VR, glasses and holograms
30:31 Can AI crack hardware and manufacturing?
32:36 Autonomous factories and reindustrialization
35:12 Do agents actually need computers?
36:23 When AI becomes alien at using computers
38:40 How Orgo plans to scale into enterprise

Casey Handmer is the founder and CEO of Terraform Industries, a company building machines that turn sunlight and air into synthetic fuel.

00:00 Making fuel from sunlight and air
02:48 How Terraform's technology works
05:54 The hardest manufacturing problems
08:59 Inside Terraform Industries
12:05 How Casey hires exceptional people
14:56 Leadership and staying focused
18:07 Developing talent and building culture
20:58 Terraform's long-term vision
27:37 What happens when energy becomes abundant?
30:28 Bringing industrial production back locally
31:56 China's energy dependence
33:49 Why Terraform may be difficult to copy
37:11 The challenge of scaling hardware
40:12 Why simpler machines win
44:08 AI and automation in manufacturing
49:32 Will humanoid robots take over factories?

Rolf Schrömgens co-founded trivago and helped build it into a multi-billion-dollar company.

We talk about raising capital, keeping control, hiring, company culture, leadership, bureaucracy, and the lessons he learned building trivago.

03:10 Lessons From His First Startup
05:51 Raising Capital: Angels vs. Institutional Investors
08:45 Why Founders Need to Maintain Control
11:48 Profitability and Choosing the Right Ventures
15:00 How Bureaucracy Kills Agility
17:48 The Myth of Hiring “Amazing People”
21:02 Building Company Culture and Values
23:56 How Leadership Changes as a Company Scales
35:35 Learning to Let Go
37:00 Becoming a System Creator

Pixar co-founder Alvy Ray Smith shares the story of trying to get Pixar funded, hearing no and making a Hail Mary call to Steve Jobs.

We talk about seeing color graphics for the first time at Xerox PARC, the early days of computer animation, George Lucas and what it took to turn the technology into a company. Alvy also explains the relationship with Disney and why protecting Pixar’s culture mattered as much as the technology itself.

Chapters:
00:00 Seeing Color Graphics for the First Time
03:02 Xerox PARC and the Future of Computing
05:57 What Made Xerox PARC Different
09:02 The Journey to New York Tech
11:59 The Early Days of Computer Animation
14:52 How Pixar Began
18:06 The Challenges of Starting Pixar
20:58 Finding a Way Forward
34:04 The Struggle to Get Pixar Funded
38:54 The Hail Mary Call to Steve Jobs
42:56 Building Pixar and Keeping It Alive
52:11 The Disney Partnership
58:02 Protecting Pixar’s Creative Culture