Coinbase CEO's Top 3 Crypto Trends for 2026 + More from Davos!

Brian Armstrong (Coinbase) – World Economic Forum & Crypto Regulation

• Legislative Breakthrough: The U.S. "Genius Act" (stablecoin bill) has passed into law, mandating that U.S. regulated stablecoins maintain 100% reserves in short-term U.S. treasuries (max 30-day maturity) with regular audits.
• Banking Partnerships: Five of the top 20 Global Systemically Important Banks (G-SIBs) now utilize Coinbase infrastructure; top-tier bank CEOs cite crypto integration as a "number one priority" and "existential" necessity to avoid obsolescence.
• Regulatory Shift: Armstrong credits the Trump administration (47th) for actively courting the industry and seeking clear rules, contrasting this with the Biden administration (46th), which he characterizes as having attempted to "unlawfully kill" the industry through inaction.
• Stablecoin Competition: Coinbase supports multiple stablecoins (USDC, Tether, PayPal) without exclusive partnerships; however, Tether currently lacks U.S. compliance with the Genius Act's 100% reserve requirements, creating a bifurcated market between compliant and non-compliant versions.
• Revenue Model Clarification: Returns on U.S. stablecoins are structured as "rewards programs" tied to user activities (trading, payments, subscriptions) rather than direct interest on deposits to comply with the Genius Act.
• Tokenization Trends: Major financial institutions, including BlackRock, are actively moving to tokenize all funds; Armstrong predicts on-chain capital formation for private companies will democratize access for non-accredited investors and increase liquidity.
• Global Competitiveness: The U.S. aims to repatriate stablecoin capital currently held offshore to maintain competitiveness against the Chinese Central Bank Digital Currency (CBDC), which now pays interest.
• California Political Outlook: Armstrong cites California's wealth taxes, housing costs, and ineffective social spending as drivers for billionaire exodus (estimated at 20% departure, creating a negative $10B tax hole), positioning Austin as a superior environment for business retention.
• Coinbase AI Integration: The company deployed an internal "Oracle" AI agent connected to all Slack, Google Docs, and Salesforce data to assist the CEO in strategic oversight, identifying team disagreements and time-allocation inefficiencies without prior human prompting.
• Future Outlook: Armstrong predicts the convergence of AI agents and crypto, where autonomous agents require stablecoins and wallets to execute payments and contracts, necessitating "Know Your Agent" protocols.

Andrew Feldman (Cerebras Systems) – AI Hardware & Infrastructure

• Hardware Scale: Cerebras unveiled the Wafer-Scale Engine (WSE), a processor 56 times larger than an Nvidia H100/B200, containing 4 trillion transistors, capable of delivering results in seconds rather than minutes for complex inference tasks.
• AI Speed & User Retention: Feldman emphasizes that shaving milliseconds from inference latency increases user engagement; Cerebras partners with OpenAI to deliver "blistering speed" to ensure users do not abandon slow models (e.g., ChatGPT) for faster alternatives.
• Power Constraints: The primary bottleneck for AI data centers is power availability, not square footage; Cerebras's deal with OpenAI involves 750 megawatts of capacity, representing a shift in industry metrics from "chips sold" to "power delivered."
• Location Strategy: Data centers are prioritizing regions with abundant, low-cost natural gas (Texas, Wyoming, Guyana, Caribbean) or hydroelectric power (Canada, Alaska); Cerebras utilizes closed-loop water cooling to mitigate water usage concerns.
• Geopolitics: The U.S. chip dominance is critical; Feldman argues the previous administration erred by restricting chip exports to allies (UAE, KSA, Denmark), whereas the current administration is empowering allies to invest in the U.S. ecosystem to counter Chinese industrial advancement.
• Supply Chain Dynamics: A global memory shortage is causing 18-month lead times for HBM (High Bandwidth Memory), driven by hyperscalers over-ordering due to supply uncertainty; this bottleneck is expected to persist for roughly 18 months.
• China AI Gap: While China lags in high-speed chip manufacturing (due to a lack of density in the Santa Clara cluster), they have surged in open-source model development and grid modernization to support AI power needs.
• AI Employment Impact: Feldman predicts immediate job displacement in middle management due to SaaS efficiency and "AI-first" startups, with broader automation effects on physical labor (robotics, logistics) expected within 3-5 years.
• Future Infrastructure: Feldman views space-based solar power and data centers as a viable long-term (8-10 year) solution for cooling and energy, though he notes significant technical hurdles regarding communication latency and vacuum cooling.
• Nuclear & Energy: Feldman supports small modular reactors (SMRs) and nuclear expansion as essential for long-term AI energy needs, criticizing Germany's recent decision to shutter nuclear plants and import Russian gas.

Jake Lucerarian (Gecko Robotics) – Industrial Robotics & Data

• Business Focus Shift: The World Economic Forum has transitioned from "DEI/ESG" performative rhetoric to "brass tacks" business negotiations focused on ROI and infrastructure building under the new political climate.
• Defense Integration: 30% of Gecko's business now serves the Department of Defense (formerly War), utilizing robots to inspect submarine welds and manufacturing processes to accelerate U.S. naval production speeds by up to 90%.
• Energy & Infrastructure: The primary growth driver is large-scale energy and mining companies; robots are used to inspect refineries, bridges, and dams to extend asset life and prevent catastrophic failures, reducing hazardous work hours for humans.
• Data as Foundation: Gecko's competitive moat is the collection of proprietary "physical world" data sets (welding, infrastructure health) that do not exist on the public internet, which are required to train specialized AI foundation models for industrial tasks.
• ROI Philosophy: Lucerarian argues that humanoid robots for consumer tasks (folding laundry) offer low ROI; the viable market is "application layer" robotics that solve high-value industrial problems (e.g., one welder supervising 10 robots).
• Welding Automation: The company is building a "foundation model for welding," using robot sensors to gather data on weld quality, enabling automated feedback loops and future autonomous repair capabilities.
• Pittsburgh Context: Lucerarian highlights Pittsburgh's transition from traditional steel to robotics/AI, emphasizing that the next industrial revolution requires applying AI to "atoms" (physical infrastructure) rather than just "bits" (software).
• Human-in-the-Loop: While aiming for autonomy, the current roadmap relies on tele-operation and human supervision to ensure safety and accuracy in complex environments like deep-sea or high-voltage inspections.
• Talent Augmentation: Robotics allows companies to upskill non-specialists (e.g., Home Depot employees) into high-wage ($100k-$150k) robotic operators within months, addressing labor shortages in skilled trades.