Big Ideas 2026: New Infrastructure Primitives

Guy Willett argues that stablecoins functioning as narrow banks won't scale on-chain finance long-term. The next phase requires native on-chain credit origination rather than simply tokenizing off-chain assets. This approach mirrors post-2008 traditional finance where private credit funds intermediate between banks and borrowers, but with dramatically reduced back office costs (1-3% annually) and greater DeFi composability.

• •On-chain credit origination can reduce loan servicing costs by 1-3% annually compared to traditional off-chain origination then tokenization
• •Market structure emerging similar to post-Basel III: stablecoins lending to curators/asset managers who make loans to end users
• •Native on-chain origination enables greater composability between DeFi projects versus copying off-chain assets
• •Private credit funds like Apollo already experimenting with on-chain credit facilitation through platforms like Morpho

Rather than tokenizing existing assets, creating synthetic representations (perpetual futures) of traditional assets scales more effectively on-chain. Synthetic dollars backed by structured products like cash-and-carry trades, infrastructure financing (GPUs, solar, batteries), or emerging market derivatives offer higher yields while maintaining dollar pegs. This 'PERPification' approach is particularly valuable for emerging markets where derivatives already trade higher volume than underlying spot assets.

• •Synthetic dollars can be collateralized by diverse assets: basis trades, currency arbitrage, physical infrastructure like GPUs/solar panels
• •Emerging market equities (e.g., Indian markets) show strong product-market fit where zero-day options trade more volume than spot
• •PERPification (creating perpetual futures) more scalable than literal tokenization of traditional assets
• •Opportunity for 'synthetic dollar factory' to package trading strategies as dollar-pegged tokens, similar to how ETFs packaged strategies in traditional markets
• •Crypto-native approach provides back-office efficiency benefits regardless of target audience (crypto-native or traditional)

Oliver Xu explains how combining AI reasoning capabilities with lab automation creates collaborative systems between scientists, AI, and robots. The near-term focus is human-AI collaboration with strong interpretability requirements, not fully autonomous closed-loop science. Progress depends on advances across mathematical reasoning, physical reasoning, simulation, and robot learning, with adoption driven by markets with established demand for research outputs like pharma, chemicals, and materials science.

• •Key innovation is combining AI reasoning/experiment planning with physical lab automation, not automation alone
• •Interpretability critical for research applications - scientists need to understand why AI systems plan experiments specific ways
• •Near-term: collaborative systems across life sciences, chemicals, materials science; long-term: fully self-driving closed-loop labs
• •Adoption fastest in markets with mature demand: pharma, chemicals, materials (vs. basic research with unclear buyers)
• •Examples include Periodic Labs, Medra (life sciences), Chemify and Yoneda Labs (chemistry), plus government initiatives like DOE's Genesis mission
• •Progress uneven across required capabilities: mathematical reasoning, physical reasoning, world models, robot learning must all advance

James DeCosta presents the greenfield strategy where AI-native startups sell to other AI-native startups at formation, then grow alongside them. This approach wins distribution before incumbents can innovate by targeting customers with few stakeholders, no switching costs, and no existing solutions. Accelerators like Y Combinator provide constant customer flow, while graduation moments (e.g., QuickBooks to NetSuite transitions) offer expansion opportunities. AI enables collapsing multiple traditional categories into single products.

• •Startups at formation have fewer stakeholders (just CEO + founders), zero switching costs, and don't need complete solutions
• •Stripe's original playbook: serve startups at formation, grow with them, use as case studies to sell to enterprises
• •Incumbents can't compete effectively due to P&L constraints - new startups represent little revenue but high service costs
• •Accelerators (YC, Speedrun, Entrepreneur First) provide constant new customer flow - Mercury works with 50% of every YC batch
• •Must rapidly ship features to reduce churn as customers grow; graduation moments offer expansion opportunities
• •AI enables category collapse: next CRM could integrate lead finding, tracking, and customer success in single platform