In this interview, Ian from TechTechPotato chats with Jon from Asianometry, a prominent tech YouTuber specializing in semiconductors. The discussion delves into a wide range of topics, including the unexpected success of Jon's content on water infrastructure, the immense water and power demands of AI data centers, and the environmental concerns surrounding PFAS "forever chemicals" in chip manufacturing. They explore the geopolitical landscape impacting global semiconductor production, specifically TSMC's international expansion and its workforce challenges. A significant portion of the conversation focuses on the transformative role of AI in content creation, from script optimization to voice generation, alongside the challenges of AI "hallucinations" and maintaining authenticity online. The interview concludes with a critical look at the struggle to attract new talent to the hardware industry, proposing innovative ideas like a chip design video game, and touching upon the future of humanoid robots and 6G technology.
Semiconductors and YouTube: A Chat with Jon from Asianometry
0:30 Jon's Unique Journey: Jon shares his unexpected path from being a "random guy" to a known semiconductor expert on YouTube, building a channel with hundreds of videos. He expresses a preference for anonymity despite his public content.
1:19 Passion for Water Infrastructure: Ian and Jon discuss the surprising success of Jon's videos on water, a topic he deeply enjoys researching and creating content on, highlighting that universal experiences resonate with audiences.
3:11 Water Demands of AI Data Centers: Jon is researching the significant water and cooling requirements of new AI data centers, noting that 60-70% of the focus is on water. They briefly touch on immersion cooling technologies, their high fluid costs, weight, and deployment challenges, as well as the industry's struggle to find widespread customers.
6:36 PFAS "Forever Chemicals" in Semiconductors: They discuss the use of PFAS chemicals in semiconductor manufacturing (e.g., photoresist, cleaning) and the associated environmental concerns due to their persistence. The discussion references IMEC (ITF World) as a key research consortium working on roadmaps for next-generation transistors and finding alternatives to PFAS, acknowledging the significant technical and financial challenges.
13:20 Geopolitical Impact on Semiconductors: Conversations with Taiwanese colleagues indicate a practical, "wait and see" approach to geopolitical shifts, despite the underlying concerns. TSMC's global expansion is seen as a necessary move due to limitations in manpower and energy in Taiwan, with efforts to tap into global talent.
15:45 TSMC's International Expansion: The successful tape-out of AMD's N4 process in TSMC's Arizona fab is highlighted, indicating better progress than anticipated for their first leading-edge international fab. Workforce retention and recruitment strategies are being adjusted in the US. The concentration of semiconductor activity in Phoenix (TSMC, Intel, Samsung) and the return of Semicon West to Arizona are noted.
22:16 AI's Role in Content Creation: Ian extensively uses AI for script optimization, finding it delivers punchier, more concise content, and has even developed a "TSMC style" memory for his GPT. AI is viewed as an "engineer having 50 junior engineers," transforming industries beyond content creation into areas like weather prediction, self-driving, and factory monitoring.
27:27 Distrust and Hallucinations in AI: They acknowledge public distrust of AI and the challenge of AI "hallucinations," where models generate incorrect information. Both Ian and Jon describe having to fact-check AI outputs, especially historical or technical details, emphasizing the need for users to understand when AI is wrong.
38:56 AI Voice Generation & Authenticity: The increasing use of AI voices in YouTube shorts and automated content is discussed, leading to a wave of "faceless YouTubers." While AI voices are improving (e.g., Eleven Labs), challenges remain with intonation and consistency, and the creators prioritize maintaining authentic human presence, including their natural vocal changes over time.
46:09 The "Deer VTuber" & Online Presence: Jon humorously mentions suggestions for him to adopt a "deer VTuber" avatar, preferring it over a physical deer head. They reflect on how VTubers and streamers use avatars and voice filters to create ageless personas, contrasting with human aging. The enduring value of honest, accurate, human-led content is emphasized.
51:03 Lack of Young Tech Talent: They express concern over the dwindling number of young, inquisitive individuals entering tech journalism and media, suggesting AI might be a contributing factor by devaluing early-career work. The industry's shift away from nurturing young writers and providing mentorship is lamented.
57:37 Attracting Hardware Talent: They identify key barriers to attracting talent to the semiconductor industry:
Compensation: Software companies offer significantly higher starting salaries, making hardware less attractive.
Perceived Difficulty: Hardware is seen as harder to get into compared to increasingly accessible software and AI.
Access to Leading Edge Technology: University students often only get access to older process nodes, making chip design less exciting.
1:07:02 Chip Design Video Game Idea: To inspire future talent, Ian proposes a "chip design game" akin to PC Building Simulator, built on AI, to allow players to simulate designing chips and see their performance.
1:07:46 Lack of Interaction with Semiconductors: They note that modern semiconductors are often invisible, making it harder for younger generations to develop the same fascination their parents might have had with visible analog chip designs.
1:08:53 The Future of 6G: They briefly discuss 6G, with the consensus that it will primarily focus on fixing existing 5G issues (bandwidth, latency, multi-user support) rather than introducing fundamentally new concepts, as 5G penetration has been slower than expected.
1:10:18 Importance of Networking: They conclude by emphasizing the critical role of networking ("who you know") in the semiconductor industry for career advancement, problem-solving, and collaboration.