“Agents are becoming more relevant to the internet than humans are.”
That single sentence from Charles Hoskinson at Consensus Miami 2026 on Wednesday distills a thesis that has Google, Amazon, and Facebook scrambling to rewrite their playbooks. The Cardano founder and Input Output CEO projected that by 2035, the majority of searches, commerce, and general activity on the internet will be conducted by AI agents rather than people. If he’s right, the advertising-driven empires of Silicon Valley face an existential reckoning.
Hoskinson didn’t mince words about how Big Tech is responding to the threat. “Amazon, Google, Facebook, they’re terrified of the agentic revolution,” he told the Miami audience. The companies are pouring capital into AI research not because they want to lead the charge, he argued, but because “all of their business models are going to be disrupted.”
The core problem is embarrassingly simple: AI agents don’t click ads. They have no brand loyalty, no emotional attachment to a logo, no susceptibility to influencer endorsements. An agent tasked with finding the cheapest flight doesn’t pause to admire the banner ad for a competing airline. It executes the query, compares prices, and books the ticket. This behavioral difference threatens the revenue models that made Google and Meta worth trillions.
The x402 Question and Crypto’s New User Base
Hoskinson posed a pointed question to his audience: “Why do you think Google is interested in x402?” The reference was to a Coinbase-backed protocol designed to let AI agents and applications make direct, programmatic payments over the internet using stablecoins and crypto rails. If agents can’t be monetized through ads, they can be monetized through transaction fees. The search giant’s interest in x402 suggests it’s already hedging against the ad-apocalypse scenario.
This prediction aligns with a growing chorus from crypto’s executive class. Coinbase CEO Brian Armstrong has said “very soon there are going to be more AI agents than humans making transactions.” Binance founder Changpeng Zhao went further, predicting agents “will make one million times more payments than humans.” Whether the CZ multiplier is hyperbole or prophecy, the directional bet is consistent: machines will dominate transaction volume.
The implications for cryptocurrency are significant, and Hoskinson views them as overwhelmingly positive. AI agents, he argued, are the “single best thing to ever happen to cryptocurrencies” because they simplify user experience. An agent doesn’t need a tutorial on MetaMask. It doesn’t fat-finger a wallet address. It doesn’t panic-sell at 3 AM because a tweet went viral.
This view echoes Alchemy CEO Nikil Viswanathan’s argument that traditional finance was built around human constraints that AI agents don’t share. Agents operate 24/7, process information instantaneously, and have no emotional attachment to round numbers. Crypto, with its programmable smart contracts and permissionless access, becomes their native financial layer in a way that legacy banking infrastructure cannot match.
Hoskinson predicted that AI will increasingly handle crypto-specific tasks: due diligence on new projects, transaction execution, and interaction with decentralized finance protocols. The mental image is an agent stack managing a DeFi portfolio, rebalancing across yield opportunities, executing swaps, and monitoring smart contract risks while its human owner sleeps. Whether this future arrives in five years or fifteen, the architectural advantages crypto offers to autonomous software are hard to dispute.
Self-Custody, Fragmentation, and the Eleven Million Token Problem
The Cardano founder pivoted from prediction to admonition, warning users against outsourcing their sovereignty to intermediaries. “You have to own your data. You have to own your identity. You have to own your money,” he said. The line was a callback to crypto’s founding ethos, delivered with evident frustration at how far the industry has drifted.
Hoskinson argued that users are “outsourcing that to custodial wallets, permissioned networks, and third parties that they come to regret trusting when they get their account shut down.” The critique lands harder in 2026 than it might have in 2019. The intervening years brought FTX’s collapse, multiple exchange hacks, and regulatory actions that froze user funds across several platforms. The Celsius and Voyager bankruptcies left depositors waiting years for partial recoveries. The track record of trusting third parties with crypto has not been inspiring.
Yet the alternative, full self-custody, remains daunting for average users. Hoskinson acknowledged this, describing current crypto onboarding processes as “complex and prone to error.” His rhetorical question cut to the point: “Is this like a product you want to use?”
The answer, for most people, is no. Seed phrase management, gas fee estimation, network selection, and bridge navigation create friction that pushes users toward custodial solutions despite the risks. Hoskinson pointed to account abstraction and chain abstraction as technologies that could simplify interaction while maintaining user control. The idea is to let smart contracts handle the ugly plumbing, so users interact with a clean interface that doesn’t require understanding which chain they’re on or what token they’re paying gas in.
Beyond user experience, Hoskinson criticized fragmentation across blockchain ecosystems as a brake on progress. “There’s been 11 million tokens issued over the years. We have enough of them,” he said. The number is striking if roughly accurate, implying the industry has created more tokens than there are publicly traded stocks globally by a factor of about 100. Most of these tokens, of course, are effectively dead, meme experiments, or outright scams. But the proliferation represents developer energy scattered across incompatible standards and siloed liquidity.
“What I want is cooperation. What I want is the mission to be achieved,” Hoskinson said. The statement reads as both a lament and a challenge. The mission, presumably decentralized and censorship-resistant finance, gets obscured when every project prioritizes its own token economics over interoperability.
Institutional Shifts and the JPMorgan Reversal
Hoskinson highlighted changing attitudes among traditional financial institutions, using JPMorgan as his prime exhibit. “Back when we started JPMorgan was turning people’s bank accounts off and now they have a blockchain product,” he said.
The reversal is well documented. In the early 2010s, JPMorgan CEO Jamie Dimon repeatedly dismissed Bitcoin as a fraud and threatened to fire any employee trading it. By 2019, the bank launched JPM Coin for institutional settlements. Today it operates Onyx, a blockchain unit handling billions in daily transactions. The pivot from hostility to product development mirrors a broader institutional embrace that accelerated after Bitcoin recently pushed toward $77,400 on strong Big Tech earnings, though ETF outflows and macro headwinds keep traders cautious about the $80,000 resistance level.
Hoskinson’s point wasn’t to praise JPMorgan. It was to illustrate that institutions follow incentives, and the incentives now point toward blockchain adoption rather than obstruction. When the technology offers settlement finality, programmability, and global reach that legacy systems can’t match, banks eventually capitulate regardless of what their CEOs said on CNBC a decade ago.
The institutional shift extends beyond banking. Asset managers, pension funds, and sovereign wealth vehicles have entered crypto through ETF products and direct custody arrangements. The SEC’s approval of spot Bitcoin ETFs in January 2024 opened the floodgates, and inflows since then have reshaped market structure. Our ETF flows explainer covers the mechanics of how these products work and why their buying pressure differs from exchange retail activity.
For Hoskinson, the institutional embrace validates the technology but doesn’t validate every business practice in crypto. His warnings about custody and fragmentation apply equally to institutions that park assets with third-party custodians without understanding the counterparty risks. The FTX collapse caught several institutional allocators flat-footed, and the lesson, verify rather than trust, applies at every asset scale.
What the Agent Future Actually Looks Like
If Hoskinson’s 2035 timeline holds, the internet in nine years will operate on fundamentally different economics. Consider the flow of a simple consumer transaction today: a user searches Google, sees ads, clicks a sponsored result, lands on Amazon, sees more ads, and eventually purchases a product while generating data for both platforms. Each step involves attention capture and behavioral nudging optimized over decades.
In the agent-dominated alternative, a user instructs their AI to find and purchase the best wireless earbuds under $100. The agent queries product databases directly, compares specifications, reads reviews programmatically, and executes the purchase via something like x402, paying with stablecoins or crypto rails. No search ad. No product placement. No attention economy. The transaction becomes a pure information and payment exchange, with value flowing to whoever offers the best product rather than whoever bids highest on keywords.
This shift threatens more than advertising revenue. It threatens the entire model of attention as a monetizable resource. Social media platforms, news sites, and content creators who rely on eyeballs suddenly face an audience of agents that don’t have eyeballs. The business model of manufacturing outrage or engagement bait collapses when the consumers are emotionless software.
For crypto specifically, agents as primary users would reshape network design priorities. Transaction throughput matters more than UI aesthetics. Smart contract composability matters more than marketing narratives. Gas fee predictability matters more than community vibes. The networks that win in an agentic future are the ones that execute reliably, cheaply, and programmatically, not the ones with the best memes or the most charismatic founders.
Solana, Ethereum, and Cardano are all positioning for this future with different technical bets. Solana prioritizes raw throughput. Ethereum emphasizes composability and the largest developer ecosystem. Cardano focuses on formal verification and research-driven development. Which approach wins may depend on whether agents prioritize speed, flexibility, or provable correctness, and that question is far from settled.
The DeFi sector is perhaps best positioned to benefit from agent adoption. Automated market makers, lending protocols, and yield aggregators are already designed for programmatic interaction. An agent managing a portfolio doesn’t need Uniswap’s frontend; it needs Uniswap’s smart contracts. The frontend is for humans. The contracts are for everyone, including machines.
Hoskinson’s warning about self-custody becomes more urgent in this context. If agents control your funds but you don’t control your agents, you’ve just added an abstraction layer without solving the trust problem. The question of who holds the private keys, and who can instruct the agent, becomes the central security challenge of agentic finance.
The Timeframe Gamble
Nine years is an eternity in technology and an instant in institutional behavior. In 2016, most crypto observers would have laughed at the idea of JPMorgan running a blockchain product or BlackRock sponsoring a Bitcoin ETF. Both happened. The lesson is that timeline predictions in tech are nearly always wrong on cadence even when directionally correct.
Hoskinson’s 2035 date for agent dominance is plausible but unprovable. The infrastructure for agentic payments, specifically protocols like x402, is nascent. The legal frameworks for agent liability don’t exist. The user interfaces for agent management are primitive. Nine years is enough time to build all of this, but it’s not guaranteed.
What seems more certain is the directionality. Agents will become more common. They will handle more transactions. They will disrupt advertising models. The debate is over speed and magnitude, not trajectory. For crypto investors and builders, the implication is clear: design for machines, not just humans. The next billion users might not be users at all.
Related Reading
- What is Ethereum? Smart contracts explained
- Seed phrase security, start to finish
- AI & Crypto news
- More on Charles Hoskinson
- More on Cardano
Sources
Standard disclaimer: we cover this market, we don’t advise you on it. Crypto can and does lose value. Make your own call.
