Q2 2026 Crypto, AI and security research summary
The research was conducted over fifty interviews and across six sub-sectors — wallet, security, AI agents, DeFi, market making and network/L2 — in April and May 2026. Four observations surfaced independently in every sub-sector. Together they describe the shape of the next cycle and what recommendations to follow.
Download the PDF Read the short note
How the corpus was built
We conducted fifty interviews with founders, operators, security researchers, market makers, DeFi protocol contributors and AI-agent builders, captured as private working notes.
The sector distribution is as follows: AI agents (13), DeFi (13), security (9), wallet (6), market making (6), network/L2 (3).
Notes were normalised into a structured format with YAML frontmatter, tagged against a controlled 26-theme vocabulary fixed before tagging began, and synthesised into eight memos. This summary is built from those memos rather than from raw notes.
The four findings, with the evidence behind them
Finding 01
Stablecoins now underpin payments, DeFi and agentic commerce
Across wallet, DeFi, security, AI-agent and market-maker interviews, stablecoins are described as the B2B payments and settlement layer being independently built in parallel. ISO 20022 alignment is now an active wallet-infrastructure conversation. A security vendor closed a stablecoin compliance deal with an e-commerce platform to help customers avoid local transfer taxes. An AI-agent platform frames the macro contest as stablecoins on open networks versus the Visa network. A trading desk reports USDT trading at a 5% premium in Dubai, 4% in India and 30% in Venezuela whilst institutions transact in stablecoins but do not open wallets or use DeFi.
Finding 02
Fintechs, not banks, are the wedge customer
An L2 voice puts it directly: "L2 stacks will need to focus on fintechs as it feels like two years from banks adopting L2s themselves." Wallet-platform leaders place fintechs ahead of banks on procurement speed and willingness to expose users to yield. Security-vendor pipelines list fintech, CEX and payments platforms as active opportunities; banks appear only as POCs. A major asset manager has crossed from exploration to a live regulator submission for an on-chain RWA fund, with market-making on Ethereum. The bank-adoption thesis lags by 12–24 months. The two camps in the corpus are not arguing about direction, but about which institutional entry point matters first. The corpus is unambiguous that fintech is the focus the next 18 months.
Finding 03
AI is becoming the forcing function — on both sides
AI agents are crypto's first true B2B customer: an AI-agent infrastructure team frames it as crypto is a perfect system for agentic commerce, and wallet infrastructure providers are designing delegation primitives explicitly for agents. The same capability is industrialising attackers. One security provider reports malicious fake-sites per protected brand scaling from roughly 10 to roughly 1,000 per batch deployed. An auditing firm describes a triple compression of shrinking audit budgets, AI-assisted "vibe coding" shipping vulnerable contracts, and AI tools actively finding novel exploits — with April 2026 (Drift, Polkadot, Cowswap, KelpDAO/LayerZero and others) as the empirical anchor. The wallet primitives an agent needs are also the surface attackers exploit; the product and security roadmaps need to be aligned.
Finding 04
Consolidation is more significant than consensus admits
In both DeFi and market making, interviewees independently described a future of two or three winners per category. The anchor quote: "We are likely to end up with Morpho + Aave + one fully regulated centralised credit market provider. Most DeFi projects will not sustain." Blue-Chip DeFi Credit pipeline confirms RWA yield as the primary growth driver. In addition, market makers are moving toward full-service investment bank structures — underwriting launches, trading, treasury management, insurance, foundation services, even M&A advisory. Hyperliquid is cited across the corpus as a forcing function for operational change in the trading stack. The partners you pick in 2026 are increasingly your investment banks, not your liquidity providers. Mid-tier generalist DeFi protocols and aggregation services are the most exposed.
The contrarian view we are taking most seriously
Of the four contrarian or bearish positions in the corpus, the one we suggest founders, allocators and projects test most aggressively against their own plans is the tokenisation divergence outcome: that crypto infrastructure achieves real product-market fit while the tradable tokens accessible to retail capture less of that upside. Infra wins, tokens lose. It accounts for the observed pattern across the rest of the findings — stablecoin success without a token tailwind, RWA TVL growth without DeFi token reflexivity, and institutional flow into market-making revenue without token economics shifting. The falsifier is specific: a 2026–2027 cycle in which token prices for the named winners materially track on-chain revenue. Until that shows up, the working assumption should be that value capture and underlying business adoption are two separate questions, priced separately.
Stablecoins are the infrastructure. Fintechs are the wedge customer. AI is becoming the forcing function. The surviving protocols and venues will be a small institutional set.
What this means for you
- Enterprise & fintech buyers Lead with security, payment integration and concrete cost savings against incumbent rails. The precedent deals already exist; use them. Bank POCs matter for 2027 but they are not your 2026 revenue plan.
- L2s & protocol teams Sequence: fintech → exchange → neobank → bank. Build for agent customers, not crypto-native retail. Plan distribution through consolidating venues and IB-style market-maker desks as a strategic choice, not a procurement one.
- DeFi & market-making founders Honestly assess whether you are on a path to be one of the two or three winners in your category. If not, specialise narrowly to serve the winners — risk, curation, trading, treasury services — or plan for acquisition.
- Security founders The threat narrative is set. The differentiator is no longer detection; it is proprietary data, key management, infrastructure protection and the credibility of a human-expert layer behind the model.
- AI-agent infrastructure builders Wallets, identity and payment primitives are the wedge. Token incentives are not. Build for agent customers, not for crypto-native retail.
- Investors & allocators Underwrite the infrastructure layer, underwrite the fintech wedge, and price in the tokenisation divergence risk. The infrastructure layer is accruing real revenue while most of the token layer is not.
What the interviews could not answer
- Institutional buyers. No interviews were conducted with large institutional players or sovereign-wealth allocators. Institutional-adoption claims here are made by teams selling into institutions, not by institutions themselves.
- Non-US/EU regulatory contexts. The research did not include APAC, Gulf or LatAm perspectives, despite emerging-market stablecoin volume arguably exceeding US-licensed activity.
- Retail and end-user perspective. The research did not include retail users in the fifty interviews; inferences about retail user behaviour come from trading desks and retail applications such as wallets.
Peebles Q2 Crypto, AI and Security report · v1.0 · 20 May 2026 · 50 interviews, period 30 March – 19 May 2026. The full report is available on request.