Anthropic spent $4.1 million on federal lobbying in the first half of 2026. That is a triple from the previous cycle. The market interprets this as defensive spending. I interpret it as a capital allocation signal.
Every dollar of lobbying is a bet on regulatory outcome. The return on that bet is measured in market access, legal certainty, and competitive moat. In blockchain, we obsess over capital efficiency ratios for liquidity provision. We ignore the same ratio for compliance. That is a blind spot.
Consensus is not a feature; it is the only truth. And in Washington, consensus is negotiated, not mined.
The data comes from Issue One, a nonpartisan watchdog that tracks federal lobbying disclosures. Total tech lobbying in H1 2026 hit a record $112 million, up 8% year-over-year. Meta, Alphabet, and Microsoft dominate the spend. But the signal is in the mid-tier: AI companies and prediction market operators.
Anthropic tripled to $4.1 million. OpenAI spent $1.4 million. Nvidia added $1.8 million. The surprise is in the prediction market vertical: Kalshi disclosed $1.8 million, while Polymarket’s spend is an order of magnitude smaller. This is not a footnote. It is a structural imbalance.
Prediction markets operate in a regulatory gray zone. Kalshi is a registered CFTC exchange, so its lobbying is straightforward—hire former commissioners, attend closed-door meetings, influence rule definitions for event contracts. Polymarket operates as a decentralized protocol, so its lobbying footprint is lighter. But “lighter” is not a strategy. It is a vulnerability.
Let me ground this in protocol mechanics. In Ethereum 2.0, I spent six months reverse-engineering the Casper FFG specification. I wrote a Python simulator to test finality conditions under network partitions. The lesson: a system that relies on social consensus without cryptographic finality is fragile. Polymarket’s regulatory approach mirrors that fragility. It relies on the social consensus of “we are just code, not a business.” The CFTC does not care about the code. It cares about the outcome.
Analyze the asymmetry. Kalshi spent $1.8 million. That is roughly 0.5% of its estimated annual revenue from event contracts (I model their revenue at $300–$400 million based on disclosed trading volume of $4–5 billion with a 4–5% spread). For Polymarket, if we assume similar volume from U.S. users (they blocked the U.S. but sophisticated traders use VPNs), their lobbying spend of sub-$100,000 implies a capital efficiency ratio of less than 0.01%. That is not efficiency. That is negligence.
But there is a contrarian view. Maybe Polymarket’s small spend is a feature, not a bug. The protocol’s decentralized governance through DAO voting makes it harder for regulators to target a single entity. The team wallet is traceable, but the core infrastructure lives on immutable smart contracts. I audited similar architectures during the Uniswap V3 deep dive in 2021. I built a Capital Efficiency Calculator that quantified liquidity density across fee tiers. The output was clear: concentrated liquidity creates higher returns but also higher impermanent loss. Translating to prediction markets: concentrated regulatory risk (Kalshi) creates higher legal certainty but also higher exposure to a single regulator’s decision. Distributed risk (Polymarket) reduces the attack surface but leaves the protocol in perpetual uncertainty.
Then there is the macro view. In 2024, after the Bitcoin ETF approval, I evaluated the structural efficiency of spot ETFs versus direct custody. The conclusion: institutional adoption increased long-term hold rates by 15% due to reduced friction. The same logic applies here. If Kalshi secures CFTC approval for a broader suite of event contracts—think sports, economic indicators, even supply chain forecasts—the liquidity will migrate from Polymarket to Kalshi. Why? Because institutional capital demands legal finality. Consensus is not a feature; it is the only truth.
But here is the counter-intuitive angle. Massive lobbying can backfire. The Terra/Luna collapse taught me that circular dependencies always crater. In 2022, I led the forensic analysis of the death spiral. The culprit was not code; it was an economic assumption that LUNA would always absorb UST issuance. The lobbying asymmetry between Kalshi and Polymarket creates a different circular dependency: Kalshi’s influence grows, it lobbies for favorable rules, those rules shrink the market for permissionless protocols, Kalshi captures more volume, it lobbies more. The loop looks stable until a black swan event—a CFTC chair change, a Congressional subpoena—breaks the circuit.
Polymarket’s small lobbying spend might be a hedge against that loop. It keeps the protocol under the radar. But it also keeps it undercapitalized in influence. The risk for Polymarket is not a sudden ban. It is a slow bleed. As Kalshi introduces regulated event contracts, traders will demand lower fees and higher assurance. The unregulated protocol will become the venue for the long tail—niche markets, illegal gambling, high-volatility bets. That sounds like a feature until you realize that the long tail attracts enforcement actions.
Let me add a quantitative layer. From the Uniswap V3 analysis, I derived a rule: liquidity always flows to the venue with the lowest friction. In prediction markets, friction includes regulatory uncertainty. We can model it as a spread. The regulated venue (Kalshi) has a regulatory spread of 0–0.5% (cost of compliance). The unregulated venue (Polymarket) has a regulatory spread of 2–5% (cost of potential seizure, account freeze, or whitelist delisting). Spreads converge when regulation becomes clear. Lobbying is capital deployed to compress that spread. Kalshi is spending $1.8 million to compress its spread by 50 basis points. Polymarket is spending almost nothing, leaving its spread wide. That is a deliberate design choice, but it is not a scalable one.
Now, the AI angle. Anthropic added the Treasury Department to its lobbying target list. That signals they are worried about sanctions compliance and money transmission laws. For prediction markets, this is a canary. If AI payments face AML scrutiny, so will event contract settlements. DeFi protocols that facilitate cross-border predictions will need on-chain identity or risk being cut off from legitimate stablecoin rails. I designed a lightweight micropayment protocol for AI agents in 2025 using ZK-rollups. The architecture proved that low-latency, private settlement is feasible. But privacy is binary—either the regulator can audit, or it cannot. There is no third state. Consensus is not a feature; it is the only truth.
The takeaway is bleak. The next lobbying disclosure cycle, due in January 2027, will reveal whether Polymarket corrects its deficit. If it does not, the prediction market layer will bifurcate into a regulated oligopoly and a permissionless sandbox. The regulated oligopoly will capture 80% of volume. The sandbox will become a honeypot for enforcement.
I have seen this playbook before. In the Bitcoin ETF review, the market assumed that approval would be a rising tide. It was not. The ETFs sucked liquidity away from self-custody solutions. The same will happen here. Kalshi’s lobbying is a capital investment that will yield a regulatory moat. Polymarket’s code is a capital investment that yields a technical moat. In the short term, code wins. In the long term, policy wins.
Death spirals are not unique to Terra. They are the natural state of any system that assumes regulatory goodwill without financing it. The question for prediction markets is not whether you believe in decentralization. It is whether you can afford to defend it.