Charles Hoskinson sat down with host Wendo on the O Show and spent over an hour going through things most crypto founders will not say in public. Midnight Network’s mainnet. Why zero-knowledge proofs are the trust engine for AI agents, not just a privacy feature. What is structurally broken inside the Clarity Act. What XRP holders are actually holding and why it is not what they think. And why the United States government cannot fix itself without a constitutional convention.
The interview was wide. The candor was unusual. This covers all of it.
Cardano’s budget season opened the conversation. On-chain governance means every funding proposal goes to a community vote, and Hoskinson described the current cycle as spicy. Tim Draper’s fund committed capital to a dedicated Cardano application investment pool. He called it the first tier-one venture capital firm to enter the ecosystem — and said they are already funded and moving. Two weeks before that vote, the Cardano Foundation and Emurgo submitted a conference funding proposal to cover Token2049 and Consensus attendance. The community killed it. Asked for too much. That is the system working as designed.
The Leios scalability solution, Cardano’s answer to the blockchain trilemma, remains on a 2026 delivery timeline. The Cardano DeFi kernel project is also progressing. It is built to make Bitcoin DeFi and Cardano DeFi interoperable at the ledger level — not a bridge, a shared execution layer. More on what that actually means later.
Midnight Went Live With Nine Partners Nobody Expected
The Midnight guarded mainnet launched March 31, 2026, genesis block confirmed, with nine federated node operators running live infrastructure. Google Cloud, MoneyGram, Worldpay, Bullish, Pairpoint by Vodafone, eToro, AlphaTON Capital on behalf of Telegram, Blockdaemon, and Shielded Technologies. Not advisory names. Not logo placements. Running actual nodes.
“Midnight is the first public blockchain that gives the world the infrastructure it needs to come on-chain without sacrificing privacy or compliance,” Hoskinson said on X at launch.
In the O Show interview, he called the current phase a guarded era. Federated infrastructure handles core operations. Full community-driven block production comes later in 2026. At launch, 130-plus post-launch bug fixes were already queued. None critical. The team planned two to three weeks of hardening before opening things further.
MoneyGram Chief Product and Technology Officer Luke Tuttle said the company’s position is that confidential transactions can serve as verifiable proof of regulatory compliance without ever exposing user data underneath. MoneyGram runs operations in more than 200 countries. Fahmi Syed, president of the Midnight Foundation, confirmed the operator group was selected specifically for mission-critical operational track records.
The NIGHT token has what Hoskinson called near-ubiquitous liquidity across major exchanges. Trading volume hit $9 billion at the all-time high. He compared that liquidity profile at this stage to early Tether, Circle, Ethereum, and Bitcoin. The token launched on Cardano in December 2025. Three months old when mainnet went live.
Midnight is probably going to be the first privacy coin listed in Japan and South Korea. That alone, in markets with strict exchange listing requirements, signals something about how the compliance architecture was built.
Here Is What the AI Agent Problem Actually Is
Most outlets covered the node list. Nobody covered this part.
Hoskinson’s case for Midnight is not primarily about human privacy. It is about what happens when AI agents start transacting autonomously on your behalf. The framing matters because it changes what the technology is actually for.
Humans communicate with body language. We read the room. An AI agent has none of that. So when Agent A needs to know Agent B is telling the truth, the only available mechanism is mathematical proof. No handshake. No eye contact. Just cryptographic verification.
“The language of agents is proofs,” Hoskinson told Wendo on X. “Crypto is the only way for agents to work.”
He broke the problem into three layers. The control layer first. AI agents are non-deterministic systems — they hallucinate, they can be manipulated, they drift from instructions. Zero-knowledge proofs can constrain agent behavior within defined parameters and verify that constraint was honored. Without this, an autonomous agent managing real money has no verifiable boundary on what it does while you are asleep.
Then the data layer. Agents need to know a lot about a user to be useful. But the more they know, the more dangerous that knowledge becomes in the wrong hands. The paradox: how do you share without sharing? An agent can prove it knows your KYC status, age, solvency, or identity without ever exposing the underlying data. That is what zero-knowledge makes possible. The answer to the paradox.
The value layer is simpler. Stablecoins are the native currency of agent-to-agent commerce. Microtransactions, 24/7 settlement, global operation without intermediaries. Hoskinson pointed to projects like X42 as infrastructure already building in this direction.
Midnight handles all three simultaneously. The account abstraction and chain abstraction work feeds the control and data layers. The DeFi kernel and Pogan project feed the value layer. When those pieces connect, an agent can be given a trading strategy, execute it on Ethereum or Bitcoin while the user is offline, and do so with Midnight’s zero-knowledge constraints running underneath.
Agents also solve the two hardest problems in crypto UX at the same time. First problem: users should not have to understand seed phrases, wallet backups, bridging, or key management. Second problem: someone still has to do all that complex work correctly. Agents handle both. They follow security best practices trivially. Multi-signature setups, encrypted backups, NSA-grade hardening checklists. Hoskinson said he tested this himself with AI tools, running a red team versus blue team hardening exercise on a Linux computer using NSA Flask guidelines. The result was reasonably good. One to two years of iteration, he said, before it is accessible for everyday users.
“We’re all working 80 to 100 hour weeks,” he said. “If we didn’t have AI, I don’t think we could do it.”
He also wrote a 300-page book on zero-knowledge proofs in a single month. Weekends only, while running Midnight full time. AI tools made that possible. Expected publication through O’Reilly or No Starch Press within three to six months. The goal is closing the gap between technical practitioners and everyone else who needs to understand ZK but cannot read academic papers.
Privacy Is Waterproofing, Not a Light Switch
Wendo asked Hoskinson to compare Midnight’s approach with XRP Ledger’s zero-knowledge integration through Boundless, which targets private smart contracts for institutional use on a public chain.
He pushed back on the framing entirely. Privacy is not binary. His analogy was waterproofing. Arizona rainstorm versus Florida hurricane versus Hawaii. The same waterproofing fails at different thresholds depending on the volume of water. And a small crack defeats the entire system regardless of how solid everything else is. One stupid design decision in a phone and an attacker can bypass the ZK layer entirely by decrypting with your keys through a backdoor.
“If someone’s honest, they can always hack your thing,” Hoskinson said during the interview. “Your system is great for a use case.”
The honest framework for evaluating any privacy claim, in his view, starts with four questions. Who is the adversary? What are their capabilities and incentives? What use cases are specifically at risk? Is this a one-time transaction or a ten-year recurring business relationship? Those answers define the trust boundary. Anyone who shows up saying their system is better than everyone else’s and has no flaws should not be trusted.
He applied this to Midnight’s own scope. When he and his team built it, they were not thinking about privacy for one chain. The design target was privacy for XRP, Bitcoin, Solana, Cardano, Avalanche, Binance Smart Chain, and others simultaneously. The Midnight airdrop went to XRP Ledger holders. He said the tokenomics were built for collaboration, not competition. NIGHT holders generate DUST, a consumptive resource token used for transaction fees. It is never printed from nothing. More network activity creates more demand for DUST, which flows directly back to holders. Non-inflationary. Cooperative. The opposite of adversarial.
Midnight City and the Governance They Are Actually Building
Midnight City is a virtual environment. Currently running dozens to hundreds of agents in beta. The final product is designed for millions. In the O Show conversation, Hoskinson described two interface modes: isometric Stardew Valley-style navigation for the city map, and a 3D mannequin view when you click on an individual agent to see inventory, character sheet, skills, outfits, and capabilities. MC Charles is his own character in that environment.
The governance model being developed for it is what Hoskinson calls dark matter governance. ZK quadratic voting, using mechanisms similar to Semaphore and Mackie, where users vote without anyone knowing who voted for what. The counts inside the system are still accurate. You get verifiable results with anonymous participation.
Instead of a linear roadmap, they built a Frontier Map. Users stake tokens against feature nodes. Conviction and duration determine which features get funded. Users are co-authoring the product over time. That mechanism also doubles as a testing ground for a next-generation ZK governance system, which Hoskinson said will eventually apply to Midnight’s broader protocol governance as the network matures.
He was clear about the timeline. Midnight did an airdrop to seven chains. Millions of people received NIGHT whether they wanted it or not. Some will dump. Some will be adversarial. Six to twelve months to reach community equilibrium. After that, governance activates in phases — Night Force ambassadors, bi-weekly hackathons, a canonical DApp layer, and then full on-chain governance through the Midnight Foundation. He said Midnight is moving faster than Cardano, which took about eight years to reach its governance phase.
The Clarity Act Is Not Being Reported Correctly
This is the section most outlets got wrong or skipped entirely.
Hoskinson’s critique of the Clarity Act is not primarily political. He has a technical objection to the bill’s structure that he laid out clearly in the O Show interview, building on remarks he gave to CoinDesk and in a March 3 broadcast.
The core problem is a three-layer legislative stack that the bill only addresses at the statutory level. The rulemaking layer is who is in the room to write implementation rules. The industry interface layer is how the industry participates going forward, through self-regulatory organizations, advisory committees, or direct consultation. Both of those layers were left entirely to agencies with no additional budget, no new mandate, and no experience regulating a two-trillion-dollar industry.
He pointed to the CFTC specifically. The bill hands the CFTC primary authority over the crypto market. The CFTC has never regulated an industry like this. It needs to hire people. It needs funding. The bill provides neither. He compared it to creating a miniature SEC inside the CFTC with no oversight and no money to run it.
Hoskinson told CoinDesk the act could require up to 15 years of rulemaking before becoming operational. He described it as a Frankenstein’s monster. The security-by-default structure is his sharpest technical objection. Every new project launches as an investment contract asset under SEC jurisdiction. The path to CFTC commodity classification is a bureaucratic maze.
“The SEC has no incentive to ever graduate anything from being a security to a non-security,” he said during the O Show interview.
Cardano, XRP, Ethereum, Bitcoin — grandfathered in. All new American crypto innovation faces a wall at the starting line or gets pushed offshore. His phrase, from an earlier broadcast: you climbed the ladder and pulled it up behind you. The incumbents get an oligarchy. New entrants get blocked.
The weaponization risk is separate from the structural critique. If Democrats control federal agencies in 2029, the bill’s rulemaking provisions give a hostile SEC the statutory tools to do everything Gary Gensler attempted but with law behind it. He argued that is worse than no bill. A bad law is far harder to change than the absence of one.
He also raised something nobody covered. The bill has no global coordination whatsoever. Crypto assets do not live inside national borders. Europe has MiCA. Abu Dhabi and Dubai have frameworks. Singapore, Switzerland, Japan, and Korea each have their own. A properly built US framework would require treaty-level coordination with those jurisdictions, bringing regulators from all relevant nations into a workshop before any bill was drafted. No attempt was made. He called it the America-great-everybody-else-sucks approach.
What a proper process looks like, per Hoskinson, is Wyoming. He helped pass the Stem Cell Freedom Act there. Unanimous. Every Democrat and every Republican voted for it. Not a simple topic. He said the outcome came from hundreds of individual stakeholder meetings before the bill was ever introduced. Legislators understood the concerns. The concerns were accommodated. By the time it reached the governor, it was a formality. Crypto legislation, at the federal level and globally, requires the same preparation. None of it happened with the Clarity Act.
What Ripple Actually Does With XRP
Hoskinson put the XRP tokenomics argument in plain terms during the interview. Ripple gave itself somewhere between 70 and eighty percent of total supply at launch. The business model, as he described it, runs in a clear sequence. Generate institutional headlines. Price goes up. Convert XRP reserves into cash. Use the cash to acquire assets — like the prime broker Ripple bought for $1.2 billion. XRP holders own none of those acquired assets. Those belong to Ripple, a private company with its own shareholders.
No buybacks. No staking rewards. Nothing in the Ripple network that creates buy demand for the XRP token. SEC filings documented the liquidation events. Hundreds of millions to billions of dollars every year. That is what the lawsuit was originally about.
“There’s nothing in the Ripple network that creates buy demand for the XRP token. There’s nothing,” he said during the interview.
He compared the structure to EOS. Block One raised $4 billion. Declared no fiduciary obligation to the EOS network. Kept the funds, now worth roughly $11 billion in Bitcoin and Ethereum holdings. EOS went dormant. Block One moved on. Hoskinson said Ripple’s incentive architecture is the same.
He acknowledged Ripple is aggressively building a web 2.5 position. Licensed company running traditional finance infrastructure, with a companion blockchain handling settlement. That model works well. Tether runs it. Binance runs it. Binance Smart Chain has 400 million users. CZ is richer than Bill Gates on the strength of that model. But in every case, the company captures the value. Not the token holder.
NIGHT was designed to work differently. Every NIGHT holder generates DUST. More network utility means more demand for DUST. No central company converting reserves into venture acquisitions. The circular economy runs through the token. Hoskinson described it as the direct opposite of the Ripple model.
The Button That Just Works
Back to the Cardano DeFi kernel and Pogan. Hoskinson’s stated goal before the end of 2026 is to complete a full circuit. Take Bitcoin. Lend it to a stablecoin. Route the stablecoin through Realy, the lending application. Generate yield. Have the yield automatically buy more Bitcoin. Return that Bitcoin to the user.
From the user’s side, it is one button. Underneath it runs across Cardano, Midnight, Bitcoin, and the DeFi kernel simultaneously. The user does not see any of it. They click the button. Bitcoin starts generating a yield. Hoskinson compared it to staking but completely private because it runs on Midnight.
“It’s like a light switch,” he said during the interview. “You push a button and then suddenly your Bitcoin starts paying a yield like staking.”
He made a broader philosophical point connected to this. Building simple things costs more than building complicated things. He quoted the writing principle: if I had more time, I’d write you a shorter letter. His version is that building complicated things was the first phase of his career. Midnight, the button, the 60-second account creation across any chain, the agents that handle all the complexity on your behalf — that is the second phase. Expensive simplicity.
Trust Collapsed. Here Is Why That Matters for Crypto
Near the end of the interview, Hoskinson made a point that sat outside the technical discussion but connected all of it.
Society has moved from what he called a blacklist culture to a whitelist culture. Previously, the default assumption was that everything was real until proven otherwise. Now, every photo, video, and audio recording is assumed AI-generated until someone can verify it is authentic. He said a society cannot function in a permanent state of distrust. Every transaction becomes a lawsuit. Every authority is questioned. Every institution carries an assumed ulterior motive.
He used a ranching analogy. If you trust your neighbor, you shake hands over dinner and close a land deal in two months for minimal cost. If you do not trust your neighbor, two years of litigation and half a million dollars in legal fees get you the same land. Same facts. Same price. Same outcome. The only variable was trust.
Extrapolate that to society. Healthcare, voting, intellectual property, dating, business — all of it breaks down when the trust substrate is gone. Crypto and zero-knowledge proofs, in his framing, are part of what rebuilds that substrate. Not through a central authority asking for trust. Through mathematical proof that removes the need to ask.
Midnight is three months old. The second half of 2026 is when privacy DApps go live at scale. Post-quantum cryptography, BIP 360, and related topics are queued for a follow-up interview with Wendo. Hoskinson said that conversation deserves its own session starting from how quantum computers actually work.
The agents are coming. The trust layer is being built. The US government is writing a law that may take 15 years to function. Midnight went live anyway.
