January 23 2026
(from NotebookLM)
Part 1: Brian Armstrong (CEO, Coinbase) – Crypto Trends & Regulation
- Political Climate & Regulation
◦ Davos Atmosphere: Shifted from "ESG/DEI" focus to "brass tacks" business and deal-making, largely attributed to the "Trump effect".
◦ US Administration Shift: The Biden administration was viewed as unlawfully hostile to crypto. The Trump administration is seen as pro-business/crypto, creating a legal path for the industry and aiming to make the US the "crypto capital".
◦ Global Competition: China is paying interest on CBDCs (Central Bank Digital Currencies), creating pressure for the US to repatriate capital.
- Banking & Stablecoins
◦ Adoption: 5 of the top 20 global banks use Coinbase for crypto infrastructure. Some banks view crypto as an "existential" threat (similar to Amazon vs. Barnes & Noble), while others see opportunity.
◦ "The Genius Act" (Legislation): A passed law requiring US regulated stablecoins to hold 100% reserves in short-term US treasuries (max 30 days).
▪ Conflict: Banks dislike this because it eliminates the "fractional reserve" lending model for stablecoin issuers, removing the need for a banking license to hold these assets. Bank trade groups are trying to undo this act.
◦ USDC vs. Tether: Coinbase supports USDC (compliant under the Genius Act/MICA). Tether is viewed as the "Wild West" option (non-compliant in the US due to lack of audited 100% treasury reserves).
- Top 3 Crypto Trends for 2026
-
The Everything Exchange: All assets coming on-chain, including equities and prediction markets.
-
Stablecoin Payments: Specifically B2B cross-border payments. This is the biggest growth area due to speed and lower fees compared to traditional FX.
-
On-Chain Capital Formation: Tokenizing private companies and funds to democratize access. This allows "unbrokered" adults (4 billion globally) and average investors to access high-quality private market assets (unicorns) previously reserved for accredited investors.
- AI + Crypto Convergence: AI agents cannot get bank accounts; they will use crypto wallets and stablecoins to transact.
Part 2: Andrew Feldman (CEO, Cerebras Systems) – AI Hardware & Infrastructure
- The Technology: Wafer Scale Engine (WSE)
◦ Concept: A single chip the size of a wafer (56x larger than Nvidia’s B200) with 4 trillion transistors.
◦ Value Proposition: Speed (Inference). Reduces deep research query times from minutes/hours to seconds.
◦ The "Google Rule": Reducing latency by milliseconds increases user engagement; slow AI drives users away.
- Business & Partnerships
◦ OpenAI Deal: Cerebras is building cloud infrastructure for OpenAI (750-megawatt deal).
◦ Cost: On-premise systems cost ~$1.5M; Cloud inference costs range from $0.50 to several dollars per million tokens.
- Energy & Infrastructure
◦ Constraint: Data centers are now measured by power (megawatts), not square footage.
◦ Power Sources: Cheapest is Hydro, followed by Natural Gas (specifically "flare off" gas from oil mining),. Nuclear is the long-term solution (modular reactors), but is 3-5 years away.
◦ Water Usage: Misconception that AI wastes water; systems use closed-loop water cooling which is efficient and does not damage the water.
- Geopolitics & Chips
◦ US vs. China: The US is ahead in chip manufacturing (hardware), but China is ahead in open-source models and grid modernization.
◦ Policy: The Trump administration is praised for allowing chip sales to allies (UAE, Saudi Arabia), whereas the previous administration restricted them.
- Employment: Middle management is currently being wiped out by SaaS efficiency and "flattening" organizations, not yet fully by AI. However, widespread AI job displacement is inevitable.
Part 3: Jake Loosararian (CEO, Gecko Robotics) – Industrial Robotics
- The Technology
◦ Function: Wall-climbing robots with sensor arrays designed to inspect critical infrastructure (bridges, dams, refineries, submarines).
◦ "Robot Native" AI: Collecting data from the physical world where no internet dataset exists (e.g., the seam integrity of a submarine hull) to train models.
- Key Markets
◦ Defense: Represents ~30% of business. Used to speed up manufacturing of destroyers/submarines to compete with China’s production speeds.
◦ Energy: Helping power companies/refineries extend asset life and increase output.
- Philosophy: ROI over Hype
◦ Criticizes "performative" robotics (e.g., folding laundry). Focus must be on high-ROI industrial problems like preventing bridge failures or speeding up naval production.
◦ Labor: Robots are filling shortages in skilled trades (welders, inspectors) and removing humans from hazardous environments, rather than just replacing jobs.
3 Key Takeaways
-
The "Business-First" Shift at Davos: There is a distinct move away from ESG/DEI rhetoric toward "brass tacks" deal-making and ROI. This is driven by a more pro-business US political climate, with regulators and companies focusing on tangible outcomes—whether that is legislative clarity for crypto or ROI for AI implementation.
-
Infrastructure is the Bottleneck (Energy & Regulation): The growth of next-gen tech is limited by physical and legal infrastructure. For AI, the constraint is power availability (leading to interest in flare gas and nuclear),. For Crypto, the constraint has been regulatory clarity, which is now being addressed through legislation like the "Genius Act" and a more cooperative administration.
-
Data Scarcity in the Physical World: While LLMs have scraped the internet, a massive data gap exists in the physical/industrial world. Future value lies in "Robot Native" data collection (inspecting physical atoms like steel and concrete) to build models that don't currently exist, as opposed to training on existing internet text/video.
Executive Summaries
Brian Armstrong (Coinbase): The regulatory war on crypto in the US is ending, replaced by a supportive administration and new laws (the Genius Act) that legitimize stablecoins, forcing traditional banks to either partner or compete for infrastructure. The future of crypto lies in "The Everything Exchange" (bringing all assets on-chain), seamless B2B cross-border payments via stablecoins, and the convergence of Crypto and AI, where AI agents become the primary users of digital wallets.
Andrew Feldman (Cerebras): The AI race is shifting from training models to inference speed. As AI becomes integrated into every workflow, latency becomes the killer of user engagement; Cerebras solves this with massive wafer-scale chips that cut response times from minutes to seconds. Success depends on securing massive amounts of power (energy infrastructure) and maintaining US geopolitical dominance in hardware while allowing allies access to the technology.
Jake Loosararian (Gecko Robotics): Robotics must move beyond "performative" consumer demos to solving critical industrial and defense bottlenecks. By deploying robots to inspect physical infrastructure (Defense, Energy), companies can generate unique, "off-internet" datasets that train AI to solve high-ROI problems, such as accelerating naval manufacturing or preventing infrastructure collapse, effectively bridging the gap between the digital AI revolution and the physical "built world".
Comments