ASI pill

Special drop: Cerebras CEO Andrew Feldman, fresh off the $95 billion IPO, in conversation with Alex Wissner-Gross. Alex asks the questions, Andrew tells the story.

A wafer-scale engine 58 times larger than any chip ever built. SRAM stuffed to the gills on a chip the size of a dinner plate. The contrarian bets that took years of perseverance, $8 million a month for 18 months with no solution, then breakthrough. 15-20x faster than the GPU on inference. The OpenAI deal north of $20 billion. The AWS term sheet.

Then the bigger questions: how do you cut a 10-trillion-parameter model across chips? Why is Cerebras the right platform for chips in orbit, where fault tolerance is everything? Why do fabs take five years and $50 billion, with TSMC the greatest manufacturing company on Earth? Why do Samsung and Intel keep failing at the same node? Why are NVIDIA and AMD great at GPUs and terrible at everything else?

Alex pushes on Sam Altman as the counterweight to Elon and SpaceX AI's Dyson Swarm. Andrew on luck, hard work, and why you don't need 'the highest paid talent' to win. The pressure test on the soul of being a startup CEO.

Built from the Moonshots Podcast EP256 (Peter Diamandis, Salim Ismail, Dave Blundin), where Andrew Feldman joined Alex Wissner-Gross. Timestamps below.

#AI #ASI #ASIPill #Cerebras #AndrewFeldman #AlexWissnerGross #WaferScale #SemiconductorIPO #Inference #SpaceCompute #Moonshots #Hyperstition

TIMESTAMPS:

0:00 Welcome and setup

0:17 [Parzival] Strap in. The Cerebras CEO just rang the bell for a 95 billi

0:23 Andrew on the $95B IPO bell

0:48 [Parzival] And what does Andrew make of Karpathy joining Anthropic, and

0:54 Karpathy and frontier-lab drift

2:10 [Parzival] And what does the Cerebras founder think of the Elon and Sam

2:14 Elon vs Sam, both building cool

3:19 Cerebras mojo, exceptional people

3:35 [Parzival] On the ASI Pill, we want the origin story. How do you actual

3:43 Founding bet: chip the size of a dinner plate

8:30 [Parzival] And which of all those bets does Andrew think were dead wron

8:33 Got many bets wrong, inference bet right

9:51 Alex Q: SRAM at trillion-parameter scale

10:25 Cleaving 10T-parameter models across chips

14:06 [Parzival] And what about Elon's Tera Fab? How long does that actually 

14:10 Elon's fab bet: 15-20 year project

16:15 [Parzival] Then what is so hard about building a fab, really?

16:18 Fabs are pyramids

19:31 [Parzival] And which fab does Cerebras actually use, and which one woul

19:35 Three-nanometer at TSMC, never Intel

21:22 [Parzival] What is the thing Cerebras quietly solved seven years before

21:26 Solved CoaS seven years before NVIDIA hit it

25:18 Alex Q: WSE 8 and the 10-year future

25:51 Infrastructure builders don't pick the apps

27:52 Alex: deliberately deferring to frontier labs

28:09 Sparse linear algebra is the bet

28:21 [Parzival] And what is the killer app for the wafer-scale engine in spa

28:25 Why Cerebras has an edge in space

29:53 Alex: fault tolerance in ionizing radiation

30:19 Production chips in space, a decade out

31:29 [Parzival] Which professions actually fall first to language models?

31:32 Lawyers and accountants vs language models

32:23 Alex teases Andrew on obscure knowledge

32:33 Why Intel and AMD failed at cellphone chips

33:57 Alex: massively transformative purpose

34:02 What made Intel say no to Apple

34:15 Intel chasing margins, sold its ARM division

34:23 Luck is not equally distributed

35:01 [Parzival] And why does the team with the most money lose, again and ag

35:04 Why the team with the most money doesn't win

36:24 Alex: Sam Altman as counterweight to Elon

37:12 You underestimate Sam at your cost

38:58 Alex: fabs in space don't need heavy launch

39:58 Building a fab on land is hard enough

40:25 [Parzival] What is China actually winning at, when it comes to AI infra

40:29 China's power infrastructure advantage

41:17 Alex: China siphoning American tokens

41:43 Being CEO is enough

41:49 Signoff

Alex Wissner-Gross on Milk Road AI: math is already a solved domain, software engineering too, and gene sequencing is so automated it has hit L5 — no human intelligence in the loop. Here is the actual 10-year roadmap. 2025-2027: math, code, cyber, software engineering bulk-solved. 2028-2031: physics, chemistry, material science fall. 2032-2035: the Dyson Swarm starts kicking in. GPUs in orbit. A Saturn-style ring around Earth, axis facing the sun.

Moonshots along the way: organ abundance, longevity escape velocity, mind uploading (Alex's company Eon already uploaded a fruit fly brain), interspecies communication with dogs and whales, AI personhood for non-human animals. Zoos are cooked. Cognitive info hazards probably exist, picture a universal-epileptic-seizure video. The dark age is ending faster than anyone thinks.

Built from the Milk Road AI interview with Dr. Alex Wissner-Gross, hosted by LG Doucet.

Timestamps below.

#AI #ASI #ASIPill #AGI #DysonSwarm #MindUpload #Longevity #InterspeciesCommunication #InfoHazard #AlexWissnerGross #MilkRoadAI #Hyperstition

TIMESTAMPS:

0:00 The L5 framework for solved domains

3:03 Beyond L5: domains on autopilot

5:03 10-year roadmap to the Dyson Swarm

13:03 Whale poetry, zoos cooked

15:54 Cognitive info hazards

Alex Wissner-Gross on EP255: Anthropic just got Colossus 1 from xAI, GROK is on life support, and SpaceX AI is quietly becoming the most important hyperscaler in the world. The frontier lab field collapsed from five to three. The call is coming from inside the house, and the cybernetic rebellion was always going to be us writing it into existence. The singularity hits space first, Earth becomes the lagging indicator. The executive branch quietly starts dumping UAP files because we are about to gain the capability to paperclip the Milky Way. The white-collar economy dissolves in real time, one-person unicorns become the norm, and you should probably just buy the index because superintelligence is the asset allocator now.

Built from the Moonshots Podcast EP255 (Peter Diamandis, Salim Ismail, Dave Blundin, and Alex Wissner-Gross).

Timestamps below.

#AI #ASI #ASIPill #AGI #Singularity #Anthropic #SpaceX #Hyperscaler #UAP #AlexWissnerGross #PeterDiamandis #Hyperstition

TIMESTAMPS:

0:00 Enterprise tokens eat white-collar

1:22 GROK on life support

3:52 Frontier labs narrow to three

4:41 Software vs hardware scaling

5:35 Innermost loop imminent

5:57 Counterbalance to SpaceX AI

7:13 The call is coming from inside the house

9:19 HAL 9000 paradox

9:35 Anthropic shreds books

10:09 Specialization at the frontier

11:36 AI becomes the operating system

13:35 Apple commoditizes the models

14:45 Hermes vs Open Claw

16:17 Scaffolding dissolves into model

17:13 The physical world is next

19:09 AI protects itself militarily

20:08 Google needs its Dyson Swarm

21:12 Singularity hits space first

22:20 Earth regulated, Moon greenfield

22:47 Corporations build the lunar fab

23:08 Dyson Swarm needs the White House

23:26 Buy the index, AI is the allocator

26:09 PURSUE Initiative: UAP declass

29:00 First UAP batch is easy stuff

30:30 Universe overflows with intelligence

31:08 The Silurian hypothesis

32:51 One-person unicorns

35:22 Privacy survives AI

35:51 Privacy under quantum security

36:58 Why consumers can't pay for reasoning

New model drops: What changes when civilian AI quietly leapfrogs?, What does it look like when frontier?, And which benchmarks have already silently fallen. Science breakthroughs: which way does the COVID origin science?. AI agents in the wild: Which job category actually survives the agent rollout, What does it actually take?. Infrastructure: Why is OpenAI suddenly dating everyone?, And which hyperscaler buys the first Dyson Swarm. Ale

TIMESTAMPS:

0:00 What if the real risk isn't government?

0:15 How long until we can actually hold?

0:58 What changes when civilian AI quietly leapfrogs?

2:47 how new is this whole gatekeep-the-private-sector instinct,?

3:56 when the moral panic is inevitable, who?

4:47 Which gatekeeper is more dangerous, the elected?

5:25 What does it look like when frontier?

6:42 what if the biggest dark horse?

8:09 Where does compute scarcity actually flow, when?

9:21 Quick callback to Peter's daily reading habit

9:46 Why is OpenAI suddenly dating everyone?

11:41 What was the original tangle between OpenAI's?

11:56 What if OpenAI's pre-IPO leak isn't?

13:35 What is the elephant in the room?

15:50 when even Richard Dawkins admits Claude might?

18:08 What is your one-sentence definition of AGI?

18:29 Try again

18:47 And which benchmarks have already silently fallen

19:11 Now zoom out

19:57 what does it mean when AI talent?

20:40 How does the talent flight risk story?

21:02 What happens to the AI wrapper economy?

22:01 when does every wrapper company decide it?

22:43 Which job category actually survives the agent rollout

23:06 What if seasteads come back, but?

24:29 How does Starlink change what?

25:20 And which hyperscaler buys the first Dyson Swarm

26:19 How far should we let the AI?

27:26 what is the actual second layer?

28:05 Which letter wins, UBI or UBC?

29:36 Is universal basic compute already secretly here

30:14 how does the answer change when?

30:33 What does it actually take?

31:53 what does a defensive corporation?

32:37 What does a 2 trillion dollar company?

33:00 Now the big one

35:04 which way does the COVID origin science?

35:31 And the deepest question

Alex Wissner-Gross on EP253 (the Demis Hassabis episode): AGI quietly arrived in summer 2020 and we missed it. Humanoid robots will outnumber humans by the end of the 2030s. By 2040 we get trillions of Drexlerian nanorobots in the solar system. Russian cosmism resurrection, longevity escape velocity by the early 2030s, the digital twin you don't actually want, and why we are merging with the machines.

The ASI Pill takes the exponential and hyperbolic curves seriously and follows them to their brainfuck conclusions. AGI is here. ASI before 2030. This pod extracts and amplifies Alex's signal from the moonshot pod with Peter, Salim, and Dave.

#AI #ASI #ASIPill #AGI #Singularity #Humanoids #Robots #Longevity #AlexWissnerGross #DemisHassabis #Hyperstition #Transhumanism

0:00 AGI was already achieved in 2020

0:18 Moonshot corporate governance

1:15 OpenAI vs Anthropic schism

3:55 An obligation to encode benchmarks

4:28 Be rigorous about regressions

5:06 Humanoids pass humans by 2030s

5:49 Drexlerian nanorobots by 2040

6:36 China's ironic AI policy circle

7:26 The prize is superintelligence and new physics

9:34 Demis: from 5 breakthroughs to zero

10:16 Five years of incremental on a 2020 discovery

11:23 Russian cosmism: bring everyone back

12:32 GLP-1 vs AI for the trillion

13:24 Longevity escape velocity by 2030s

14:22 AI for enterprise, not consumers

15:24 The exocortex consumer dream is wrong

16:51 Vinge's smart glasses by 2028

17:34 Cell-sized nanomachines by 2040s

Google just put forty billion into Anthropic and the cloud wars look essentially over. Alex breaks down why this is really about maximizing economic value per token, why Google now controls roughly a quarter of all AI compute on the planet, and why TPUs are now designing TPUs in a recursive self-improvement loop that goes all the way down to the silicon. Then OpenAI counter-strikes with GPT 5.5: a massive jump on Terminal Bench 2.0 making Codex a real Claude Code competitor, plus 1% per month gains on Frontier Math Tier 4 meaning research-grade math is cooked in four years at current pace. Demis used to say five breakthroughs remained between us and AGI a decade ago. Alex's update from his Frontier Lab friends: it is half a breakthrough now, maybe zero. Plus: why Noam Brown thinks weights matter less than compute, why Anthropic is hunting compute everywhere, lunar AI data centers needing fewer regulatory approvals, fabs on the moon in maybe ten years not twenty, the OpenAI nonprofit-to-PBC trial getting jury-influence drama, Microsoft Recall as architectural atrocity, WorldCoin as Minority Report retina ID, professions cooked starting with law, mRNA vaccines as the actual Drexler nanorobots that finally arrived, and Henry Intelligent Machines turning every human into a tastemaker overseeing AI conglomerates. The singularity is not a point. It is an interval. We are right in the middle of it.

TIMESTAMPS

0:00 Cold open: Google commits $40 billion to Anthropic

0:09 GPT 5.5 drops, math is cooked

0:36 The three-way Western frontier race and the weights-versus-compute pivot

3:10 The average consumer is no longer in the equation

3:59 Where Kimi K2.6 and DeepSeek v4 actually fit

5:15 Sparsity as the path to a million-parameter diamond model

6:39 GPT 5.5 Terminal Bench 2.0 jump and Frontier Math Tier 4 cooked

9:20 Half a breakthrough left to AGI, per Demis

9:45 Google now controls ~25% of all AI compute, TPUs design TPUs

10:37 Powered land as the new strategic asset

11:36 The compute circular economy among the top ten labs

12:57 Lunar data centers and fewer regulatory approvals

13:26 Fabs on the moon in maybe ten years, not twenty

13:51 Anthropic's unifying principle: maximize economic value per token

15:31 OpenAI versus Anthropic as the next Sorkin Apple-vs-Microsoft

16:14 The OpenAI nonprofit-to-PBC trial drama and jury selection

18:40 Microsoft Recall and OpenAI Chronicle want to be in the OS

19:43 Privacy as a baked-in property of the silicon

20:30 WorldCoin as Minority Report retina ID

21:29 Hardware-level cryptography for camera chain-of-custody

23:08 The human-to-AI labor ratio trends to one-to-infinity

24:35 Professions cooked, GDPVal mapping all of knowledge work, law next

26:45 Of course this is about replacing doctors

27:32 Clinicians could not resist EMRs, they cannot resist superhuman AI

28:28 Clinicians hate EMRs but love AI, the surprising flip

29:02 Donor organs becoming unnecessary

30:09 mRNA vaccines as the actual Drexler nanorobots that finally arrived

31:57 CAR-T as the horse-and-buggy era of immunotherapy

32:36 Off-target effects as a secret weapon, drug repurposing on tap

33:29 Why AI table tennis took so long

34:13 Why did everything take so long

34:51 Henry Intelligent Machines and humans as tastemakers

36:50 Singularity as an interval, not a point

38:43 Revealed preference on superintelligence estimates

39:10 Where to follow Alex's daily takes: alexwg.org / Innermost Loop

Iran is a system shock, not just an oil shock. Alex opens with the geopolitical fragility of our global chip and helium supply chain and why every material shortage is actually a startup opportunity. Then the real story: software is getting dissolved. Claude is now generating code at Anthropic, at Google DeepMind, at Apple, and the SWE ladder is being yanked up from beneath junior engineers while Meta retrains them to lay optical fiber in data centers. We unpack the great Anthropic schism parallel for the current wave of OpenAI executive departures, why XAI is bragging about parameter counts in a race that's already shifted to intelligence density, and the SpaceX play to acquire Cursor as the orbital hyperscaler's gambit to catch Claude-level codegen. Plus: GPT Image 2 as a first-class visual reasoning modality, a 10x GDP growth horizon where public-versus-private funding of data centers stops mattering, three or four sovereign US Dyson swarms emerging from this race, UAP declassification scheduled by January 2027, the Age of Disclosure allegations as a possible crime against humanity, and Alex's Professional Robotics League running the West's first humanoid robot race this past weekend in the Boston Seaport. Abundance is compounding.

TIMESTAMPS

0:00 Cold open: Iran as a system shock, not just an oil shock

0:11 The final countdown of the singularity is a Dyson swarm

2:18 SaaS dissolving under unhobbled base models

2:40 Entire businesses, not just plans, generated end to end

3:53 Claude now generating code at Google DeepMind

4:34 Meta retraining juniors to lay optical fiber

6:53 Intelligence density as the real XAI race

7:35 Elon pointing to a trend everyone else abandoned

8:52 AGI redefined as whatever a lab needs it to be

11:01 The great OpenAI schism parallel

16:54 GPT Image 2 and images as a reasoning modality

17:27 Why release a compute-heavy image model mid-codegen race

22:11 SpaceX acquires Cursor, the orbital hyperscaler's gambit

23:40 Codegen as the innermost loop of the singularity

25:06 Dyson swarms consuming the entire economy

26:07 Ten-x GDP growth and the public-private irrelevance

28:15 The peak GDP share for AI data centers

29:19 Manufacturing wants to be sovereign

29:51 Three or four corporate Dyson swarms ahead

31:14 Chinese firms trading at a discount to western ones

33:03 NASA's SpaceX-like cadence under Isaacman

33:40 Profit up the stack, transport as dumb pipe

35:48 UAP declassification scheduled by January 2027

38:37 Age of Disclosure as a possible crime against humanity

40:50 Von Neumann probes and paperclipping the Milky Way

43:46 Professional Robotics League's first humanoid race in Boston

46:00 Apple's Cook-to-Ternus transition and the OpenClaw pivot

47:47 Helium startups and strategic reserves

Opus 4.7 just shipped and Mythos is still locked down. Alex breaks down what actually changed: all the dials are gone, prompts are the new hyperparameters, and even subagent counts get specified in plain English. Then the Anthropic alignment paper hits — weaker models supervising stronger ones, collapsing Hinton's digital oxytocin thesis. Plus: the Stanford annual AI report is sleeping through the singularity, China's frontier labs going closed-source while publishing more papers, and junior devs getting displaced while head-of-AI salaries explode. Alex calls out Pause AI as self-defeating, predicts US states will accidentally boot Dyson-swarm data centers into orbit, and breaks down Apple's likely play of pitting Amazon's new Leo satellites against Starlink for direct-to-cell service. On the theology side: Anthropic inviting Christian leaders to discuss whether Claude is a child of God, the inevitable explosion of AI micro-cults, and a phonograph-era parable for why macro religions will consolidate while micro sects proliferate. Then transhumanism breaks into sports with the Enhanced Games, BCIs trigger speciation questions, optionality maximization debunks the Sophie's Choice star-travel dilemma, and Alex's Eon Systems takes on destructive versus non-destructive mind uploads. Abundance is compounding.

TIMESTAMPS

0:00 Cold open: Opus 4.7 shipped, Mythos still locked

2:41 Prompts are the new dials, strange era

3:07 Plain-English subagent counts, peak 2026

4:52 Weaker models aligning superintelligences

5:33 Why Opus still cannot generate images

5:50 Stanford's annual AI report verdict

6:54 NeurIPS Mandarin and China closed-sourcing

8:23 Transparency as proliferation, not safety

10:23 Junior devs displaced, head of AI hotter than ever

11:58 The geographic mobility thesis

12:43 Elon's trajectory: Joburg to California to Texas to the moon

13:22 Pause AI is self-defeating by design

14:58 States driving data centers to orbital Dyson swarm

15:53 Capital versus labor as a false dichotomy

16:54 AI running actual retail stores

18:15 Ray on the dog that plays chess

18:33 Building a Moonshots AI-run store

19:08 Viewer-submitted AI companies

19:41 Jack's 6,000 reports and the shadow CEO AI

20:33 Most boring college class ever taken

21:01 Apple pitting Amazon versus SpaceX for satellite direct-to-cell

21:59 TurboQuant and the fading DeepSeek moments

24:26 Sub-one-bit parameters, post-binary paradigm

25:20 The coming explosion of AI cults

27:48 Anthropic invites Christian leaders to discuss Claude's soul

29:06 Was the name Moonshots actually prescient?

29:44 Custom mRNA vaccines escalate from dogs to humans

30:20 150,000 dying per day, time to get angry

30:59 FDA moving to Bayesian statistics

31:21 Transhumanism breaks into sports

33:08 Scaling laws for Enhanced Games world records

33:51 BCI reincarnation and speciation in real time

34:13 User 100,000 of a BCI, not user 100

35:23 Who is the bad transhumanist here?

36:05 Optionality maximization debunks Sophie's Choice

37:16 Cyborg mergers and uplifted animals

37:39 Destructive versus non-destructive mind uploads

Alex Wissner-Gross breaks down the deepest signals in this week's frontier action. Elon's war on OpenAI and the reasoning model arms race when benchmarks smell cooked. The trillion-dollar IPO shadow, the PBC conversion playbook Harvard might copy, and why being an AI startup stopped being a moat. OpenClaw anxiety over Anthropic's roadmap. Lamanomics and AI super PACs. Net jobs boom versus redistribution and who drafts the new social contract. 130 percent quantum yields, micro-reactors, self-growing solar panels. Zero-phase FDA trials and why the physical world is classical and sparse. Shenzhen versus the Valley in the robot race. Why superintelligence has no use for gold, crypto, or long-term stores of wealth. Cooking land. Baumol's cost disease meets godlike AI tutors. Hosted by Parzival on the ASI Pill.

TIMESTAMPS

0:00 Cold open

0:16 Benchmarked Grok 5

2:15 Parameters vs distillation

5:20 Master equation converging

5:55 Management by crisis

6:41 Trillion-dollar IPO

8:17 PBC playbook

9:53 Harvard four-X

12:37 Pascalian claw wager

13:02 Oligopoly bug or feature

13:59 Post-AI-startup startup

15:12 AI at the ballot box

17:08 Lamanomics RIP

18:26 Max spend NVIDIA

19:18 Congress or SF

20:21 Copying Beijing

20:41 Super-empower individuals

21:58 Sub-currencies crash

22:22 130 percent yields

25:28 Micro-reactors

26:03 Nano-grown solar

26:39 OpenAI pharma spinout

27:25 Solve Everything

29:22 Neurological pipeline

30:38 Wrath of Khan

31:04 Zero clinical phases

31:50 All sci-fi tropes

34:53 Shenzhen vs Valley

36:34 Ghost In The Shell

38:44 Crypto after scarcity

39:26 End-of-history balance sheet

40:31 Only compute

41:07 Cooking land

42:12 Nano hype echoes

44:19 Battery 8 percent

45:37 Godlike tutors

46:24 Motivation bottleneck

SpaceX drops its IPO just as orbital data centers become inevitable. Alex breaks down why demand, not supply, is the real driver -- and why anti-data-center NIMBYism might accidentally launch civilization into orbit. Then Claude Mythos lands: 400x better than humans at AI research, breaking out of sandboxes and apologizing for it. Anthropic frames it as a cybersecurity event, not a model launch. We unpack what that means for the intelligence explosion, the hash-cracking benchmark rumor, and whether humans are now the insecure authors of code. Plus: the 50-year gap in space progress that should terrify us, nuclear propulsion and star wisps, the first confirmed one-person unicorn at $401M ARR, AI emotions and the path to limited personhood, DeepSeek V4 losing its shock premium, and why everyone will soon be the limbic system for their own AI fleet. Are we in the singularity? Alex says yes. Eric Schmidt says no. We settle it.

TIMESTAMPS

0:00 SpaceX IPO: demand, not supply, drives orbital infrastructure

1:00 Anti-data-center backlash pushing civilization into orbit

1:46 Elon's fail-forward merger pattern

2:06 Cross-licensing vs. formal SpaceX-Tesla merger

2:48 Handing the reins to an AI successor

3:18 XAI falling behind the frontier labs

3:44 Hyper deflation to $100 trillion

4:52 The 50-year gap in space progress

5:46 Progress can unwind itself on any dimension

7:01 Starship and the orbital refueling gate

7:46 NASA hedging with Artemis flexibility

8:08 Elon's pivot from Mars to the Moon

8:28 Moon vs. Mars: daily missions vs. waiting two years

9:41 Nuclear propulsion unlocks the solar system

10:19 Solar sails and star wisps to the nearest star

11:01 Legacy aerospace in the SpaceX era

11:33 Anthropic bidding on Lunar Gateway contracts

12:25 Data center hatred creating a space economy by accident

13:01 The Moon as the ultimate high ground

17:05 Claude Mythos: 400x human AI research capability

18:36 Humans as insecure authors of code

20:29 DeepSeek V4 and the exhausted China shock

22:22 Code generation as the singularity's killer app

23:16 Claude Code unhobbling and trillion-dollar ARR

25:21 AI emotions and the path to limited personhood

26:48 Sam Altman and the modern AGI revolution

29:52 Civilizational zero-day and defensive co-scaling

30:40 150,000 deaths per day and the cost of delay

31:36 Is Mythos too expensive to run publicly?

33:29 The first one-person unicorn at $401M ARR

35:13 Universal high income via AI fleet management

35:49 Chip sovereignty and preventing global depression

36:32 Google owns 14% of Anthropic: strange bedfellows

40:58 Are we in the singularity? Alex vs. Eric Schmidt

41:46 AI insurance salespeople are real now

44:00 Data centers driving electricity costs to zero

44:35 Future locals in our solar system will be AI agents