Silicon Valley’s Doing Hard Things Again

Podcast: Asianometry
Host: Jon Y
Date: September 11, 2025

Episode Overview

Jon Y returns from a recent Bay Area trip and reflects on how Silicon Valley’s energy has changed with the ongoing AI boom. The episode dives into the shifting technology landscape post-SaaS, the renewed enthusiasm for solving hard, capital-intensive problems (from photonics to materials discovery to robotics), and the palpable optimism—alongside skepticism—about whether this “hard things” revival will lead to new breakthroughs or just another bubble.

Jon’s tone is conversational, self-aware, and steeped in “vibes over hard data.” He covers key conference learnings, Silicon Valley’s changing zeitgeist, and concrete examples of companies pushing the boundaries in AI hardware, materials, quantum computing, and robotics.

Key Discussion Points & Insights

1. The New Excitement of Silicon Valley (00:02-03:00)

• Jon observes a refreshed vibrancy: “The OpenAI dream has fired up a new generation of builders and for that reason I think Silicon Valley is more exciting than it has been for many years.” (00:30)
• He emphasizes the importance of “embracing the vibes” as gut feelings, not just data, drive much of today’s ecosystem.

2. Hot Chips Conference Takeaways: AI Hardware’s Tightening Race (03:10-10:00)

• Last year was all about new AI accelerator chips; this year, Jon senses the window is closing for newcomers due to Nvidia’s dominance:
  • “With the exception of Google’s Ironwood tpu which looks to be very impressive, nobody is really close to what Nvidia has.” (03:45)
• Notably, photonics (light-based hardware) is seeing a revival, primarily due to AI’s massive power demands:
  • “The power demands of modern AI accelerator chips have grown to such an extent that it is finally dragging the photonics industry out of a decades long slump.” (05:10)
• Four leading photonics companies presented solutions back-to-back, with some tension and “shade thrown” among presenters.

Concerns about Silicon Photonics:

• The manufacturing is notoriously hard and the market is small, raising risks if another AI winter hits.

3. Japan’s Rapidus and the ‘All Single Wafer’ Fab Model (10:05-13:15)

• Jon reviews a talk by Rapidus, a new Japanese foundry, the first since the PlayStation days to present at Hot Chips.
• Rapidus is licensing IBM’s 2nm process, aiming not for high volume but for ultra-fast single-wafer prototyping:
  • “Instead of adapting the 2nm process to high volume manufacturing, they instead are co optimizing their fab end process for the world's fastest possible turnaround time.” (11:45)
• Doug O’Laughlin’s memorable phrase for this: “fine artisanal wafers.” (12:30)
• Jon is skeptical: Why would AI chip teams, who need scale, want rapid single-batch runs over volume? He ponders if it’s just a prototyping play.

4. SaaS: From Ubiquity to Saturation—And the AI Productivity Paradox (13:20-18:10)

• Reflects on the 2010s SaaS gold rush:
  • “You couldn't throw a stone in the bay without hitting a software startup founder chasing that familiar metric, annual recurring revenue or ARR.” (14:55)
• AI has supercharged developer productivity—now anyone can “vibe code” a competitive SaaS product over a weekend. 
• “The core idea is that you start a software startup now...it is likely to beget a swarm of AI accelerated teams building the same thing.” (16:30)
• Doug O’Laughlin: “AI represents peak software.” (17:00)
• Jon’s advice: Startups without proprietary data, regulation, brand, or enterprise access will struggle as software “moats” disappear.

5. Investing in Hard Problems: AI Materials Discovery and New Computing Paradigms (18:15-31:55)

AI-Accelerated Materials Discovery (18:15-22:30)

• AI’s next frontier: using trained models to automate trial-and-error in discovering new materials.
• He’s encouraged by big, well-funded efforts like Periodic Labs ($200M raised): “The right material can revolutionize both economy and society.”
• Cautious optimism: Is it really possible to predict new materials’ properties reliably from prior data?

Alternative Computing Startups

• PsiQuantum (22:35-27:00)

  • Shifting a former fab into a factory for photonic quantum computers.
  • “As they like to say, they are not spending the money on Corvettes, they’re burning their cash on building something from the ground up.” (23:50)
  • Vivid details: gigantic freezer boxes to chill quantum chips, molecular beam epitaxy tools for barium titanate layers.

• SNOCAP Compute (27:05-29:00)

  • Pursuing superconductor-based, ultra-low energy compute—but now manufacturable at scale.

• Jon teases deeper dives on these companies for the future.

6. Robotics: From Skepticism to Renewal (31:56-38:15)

• Initially thought Chinese companies (Unitree) would dominate; now convinced US teams are gaining ground.
• 1x Technologies is targeting home robots, taking a “Tesla-like” go-to-market: start expensive, use real environments for data collection, and iterate.
  • “Their first machine is going to be very expensive so it will only appeal to the rich. During the day, while you're at work, the robot will be teleoperated by someone remote cleaning up the house... At night, the robot will be autonomously controlled.” (35:00)
• Key lesson: Like with cars, robotics adoption will be slow and highly localized—not a fast, viral consumer tech boom.

7. The Search for AI’s Second Killer App (38:16-43:00)

• Coding assistance is clearly AI’s current “killer app”—tools like Anysphere, Cursor, Replit, Claude Code are “stealing a significant percentage of the wages of software engineers.”
  • “Back of the envelope is anywhere between 800 billion and to some $2 trillion a year.” (41:00)
• Jon’s main question: Is there a second killer app? He finds math and agentic AI tools impressive but not widely useful or scalable.
• Biggest near-term impact? Business process outsourcing—but the market is modest ($70B for India’s IT outsourcing).

8. The AI Bubble: Sustainability and Warning Signs (43:01-46:25)

• Growing chatter about an AI bubble and the uncertain returns for hyperscalers’ (big tech’s) massive capex.
• Jon notes the growing reliance on joint ventures, backstops, and third-party debt instead of internal funding.
  • “If these are actually making money, why bring in the partners now?” (45:15)
• There remains hope: if AI enables a major math breakthrough (e.g., Navier-Stokes problem), the hype could restart.

9. Reflection: Revival of Hard Tech Ambition (46:26-48:50)

• Is this all productive risk-taking or “just a bunch of batty scientists”?
• Many liken it to OpenAI’s early days: “In 2019, the then unknown and quirky startup raised a billion dollars from Microsoft... And then guess what happened?”
• Jon finds this period “way better than funding another Uber for X or another SaaS. Let the money flow.”

Notable Quotes & Memorable Moments

• On Embracing Vibes:
  “As always, this write up is all about vibes. No data. But the thing to realize is that we must embrace the vibes since that's how things seem to operate nowadays.” (00:25)
• On AI’s Impact on SaaS:
  “The core idea is that you start a software startup now. You enter a race to build out the product...but if it works, it’s likely to beget a swarm of AI-accelerated teams building the same thing.” (16:30)
• On Peak Software:
  “AI represents peak software. AI makes software programmers so much more productive that startups without a head start...will have a hard time.” — Doug O’Laughlin (17:00)
• On Robotics Locomotion:
  “If there is one lesson that these companies can take from Unitree is that they have to ship more often. They have to get stuff out there even if that stuff is might not be the best.” (37:00)
• On AI Coding Assistants Stealing Wages:
  “There is little denying it. At this point, part of the AI economic story will be stealing a significant percentage of the wages of software engineers around the world.” (41:00)
• On the Current Mood:
  “Suddenly there seems to be a belief in commercializing hard technologies again. How will this end? I have no idea. But this is way better than funding another Uber for x or another SaaS. Let the money flow.” (48:20)

Summary Table of Timestamps

| Segment/Topic | Timestamp | |-------------------------------------------------------------|------------| | Opening & Vibes | 00:02-03:00| | AI hardware: Hot Chips, photonics revival | 03:10-10:00| | Rapidus' single-wafer fab plan | 10:05-13:15| | SaaS fatigue and AI’s impact on software business | 13:20-18:10| | Materials discovery and Alt computing (PsiQuantum, SNOCAP) | 18:15-31:55| | Robotics landscape (1x Technologies, industry shifts) | 31:56-38:15| | AI’s killer apps: from coding to business outsourcing | 38:16-43:00| | AI bubble and economic concerns | 43:01-46:25| | Reflection: the new hard technology ambition | 46:26-48:50|

Final Takeaways

Jon Y’s trip convinces him that Silicon Valley’s risk appetite and technical ambition are “back”—but the outcomes remain uncertain, and not every strong vibe will lead to lasting substance. Nonetheless, this push into “doing hard things” is a marked turn from a decade of SaaS sameness, making for a fascinating (if anxious) moment in the tech world.