hopium report: AI wants to date you (and buy your bags)

good morrow,

reporting live from this week's psychotic break, where reality has decided to outperform our finest satire. this week, a french madame fell into a parasocial relationship with an AI brad pitt (honestly, who hasn't?), the world's most powerful nation is preparing for an inaugural crypto ball, and tiktok's impending ban has inspired gen z to hard pivot from learning dance trends to conjugating chinese verbs.

Duolingo

@duolingo

oh so NOW you’re learning mandarin

enjoy,

xx c

MoonPay Acquired Helio for $175M. The $ANIME token, backed by Azuki, is set to launch on Eth and Arbitrum this month. Microstrategy bought another 2,530 Bitcoin for $243M. Trump expected to issue executive orders related to crypto policies on his first day in office. Cookie.fun launched an AI agents index and in-platform swaps. Sony has officially launched its Layer 2 platform, Soneium. Punks market moving after sale rumors. Virtuals announced an update to their value accrual mechanism.

We've been seeing interesting activity on Rodeo, a new minting platform that feels more like an "Instagram with economics." Rodeo seems to be driving more organic social collection behavior - when we post something, we're seeing our actual friends collect it. While the raw numbers are still much smaller than Zora (in terms of daily active users and mints), what's interesting is the distribution pattern. Instead of just a few whales, we're seeing more long-tail activity with most posts getting between 5-100 collects.

Hallucination Yield

We've been fascinated by goodalexander's insights about what he calls "hallucination yield" - the difference between a project's actual market cap and what AI language models think it should be worth. He tested this by asking ChatGPT about Virtuals, and the AI consistently valued it at $7-12 billion back when it was worth almost nothing. He also applied his background in hedge fund marketing analytics to compare AI's affinity for different memecoins, finding early signals that Pepe would outperform others. What's particularly interesting is how different regional AI models show distinct biases - Western models favor ESG and safety-aligned projects, while Chinese models like Qwen demonstrate different valuation patterns. It's also interesting to note how these AI models seem to have a deep affinity for tokens that dominated early internet forums - like Bitcoin, XRP, and LINK. This makes sense since these models were trained on historical internet data. So many top 100 projects are simply there because "they've always been there," and this historical presence seems to translate into both mindshare and practical liquidity advantages.

As AI democratizes analysis, traditional analytical advantages are disappearing. Instead of competing with institutional AI systems, Alex argues the future belongs to those who can orchestrate human intelligence in ways machines can't replicate. This insight led him to build Post-Fiat on XRP, focusing on using AI to improve existing crypto infrastructure rather than building entirely new infra.

His framework for surviving in AI-driven markets emphasizes systematic integration of AI into decision-making, understanding model biases, and building systems that enhance rather than replace human agency. He suggests a practical approach: recording market observations with tools like Otter, analyzing transcripts with various AI models, and creating platforms that aggregate human intelligence in unique ways.

Institutions vs AI: Tyler Cowen on Why Progress Will Stay Slow

In a recent interview, Tyler Cowen gives a notably modest view on AI's economic impact: he expects AI to boost growth by around 0.5% annually over decades - meaningful but not explosive. He argues against predictions of massive 20%+ growth rates, pointing to persistent "bottlenecks" in areas like gov't, healthcare, and education that make up roughly half the economy and are slow to change. In his view, humans, regulations, and institutional decision-making processes will remain key limiting factors.

On the talent and progress front, Cowen pushed back against narratives of declining competence, arguing instead that we're seeing increasing variance - the top performers are getting better while a thick middle band is declining. He views the Bay Area as having the world's most ambitious and intelligent people, but suggests this leads to overvaluing intelligence relative to other factors. His core concern about progress isn't about talent or technology, but rather how new technologies historically become instruments of war, potentially leading to more destructive conflicts even if they become rarer.

Dwarkesh Patel

@dwarkesh_sp

I hassled @tylercowen about why he doesn't expect explosive economic growth from AGI. How could we possibly add 100 billion extra workers and only get 0.5% more growth?

Also featuring Stalin's library, EU decels, and how Churchill was an underachiever.

Hilarious and provocative…

Post-Regulatory Clarity: Crypto's Unbounded Vision

Crypto CliffsNotes

Crypto has grown larger than you think. Stablecoins, Prediction Markets, DeFi, RWA, DePIN, Decentralized AI, Gaming, Memecoins, DeSci, Social

https://medium.com

Stephen McKeon, a former finance professor turned VC at Collab+Currency, provides a comprehensive CliffsNotes overview of crypto's practical applications in 2025. After deploying $185M across 170 crypto projects over seven years, he frames the key use cases driving institutional adoption.

His post asking "What is crypto good for?" is a timely reminder of the industry's progress amidst market volatility. What's notable is that stablecoins have emerged as the killer app, processing $5.6T in economic activity plus $20T in internal transfers in 2024 - surpassing Visa's $13.2T network. Tether expects $10B in net income for 2024, matching tradfi giants like Goldman Sachs. Major players including PayPal, Sony, and Deutsche Bank are launching their own stablecoins, while Visa is offering "stablecoins-as-a-service." The largest crypto acquisition to date was Bridge, a stablecoin platform (~$1.1B). The piece makes a compelling case for DeFi's growth, with decentralized exchanges processing $2.5T in spot volume over the past year and lending protocols holding $50B in collateral. But perhaps most interesting is how tradfi assets are moving onchain - tokenized treasuries grew from $105M to $4B in just two years, with BlackRock and Franklin Templeton entering in 2024. The driver? Near-instant settlement could unlock $1.4T currently tied up in settlement reserves.

Things that seemed unclear before, like PayPal's stablecoin launch, now feel obvious in retrospect. This connects interestingly back to PayPal's early days, where even in the late '90s they were thinking about the future of internet finance, though they couldn't have imagined crypto specifically. We're also seeing a significant shift in how founders approach building in crypto - there's a sense that the era of regulatory uncertainty might be ending, leading to what we expect could be a "tsunami" of new launches. One of the most interesting insights is the second-order effects: it wasn't just that teams weren't launching tokens, but many had "watered down" their original visions to fit within regulatory grey areas. Many founders are now asking themselves, "Where have I Gensler'd my thinking?" and are considering how to push their visions to their full potential without those constraints.

Beyond Trading Bots

We explored some interesting developments around AI and wellness. First, Fred Wilson's prediction about an AI doctor with "Mr. Rogers' personality" treating millions at zero cost by 2025. Which led to an intriguing concept called "Tesla for Souls" - based on a four-year-old post about a founder who wanted to invest his exit money in increasing happiness. We're seeing this concept being implemented now with a bounty system where people offer significant money (like $10,000) for specific outcomes like reducing anxiety or becoming more assertive.

This model could be applied broadly across knowledge-based sectors. Instead of traditional coaching or consulting, you might pay an amount of money to get the combined expertise of three top experts in any field, made infinitely scalable through AI agents. This could apply to crypto, investing, or any specialized domain - it's not about generally available data, but "hyper-specific contextual data" bundled together for specific problems.

Live & online Fridays at 10am PT / 1pm ET.

Jesse Walden

@jessewldn

users → owners: why does it matter?

5 things you can now own onchain

1/ money: self-custody enables faster/cheaper tx, more competition for services, global access

2/ products: participate in the economic upside of things you contribute to

3/ identity/data: conveniently…

Mason Nystrom

@masonnystrom

New Blockchain Letter just dropped.

Three Big Trends in DeFi:

Consumer DeFi – fintech-grade UX meets onchain yields

Real World Assets (RWAs) – the bridge to TradFi capital

DeFi as Platforms – evolving from standalone apps to DeFi platforms

Let’s break it down

Dan Morehead

@dan_pantera

Our latest Blockchain Letter uncovers the opportunities and trends shaping crypto in 2025.

– The Path to Adoption: Blockchain’s Next 100x Opportunity
– Predictions for Crypto in 2025
– Crypto: The Ironic Answer to De-Dollarization
– Three Trends in DeFi
panteracapital.com/blockchain-let…

The Fourth Way

@The4thWayYT

bro if u lead me into fucking temptation instead of delivering me from evil i fucking swear

smilinglllama

@smilinglllama

Ethereum.org

@ethereum

hello world computer

mom

@pencesmomwife

incredible things happening on xiaohongshu

The times they are ai-changin'

Two profound things are happening right now for entrepreneurs. First, AI is creating an entirely new set of capabilities for people to build new experiences and solve new problems. In the same way mobile and the cloud unlocked a new canvas for application developers, AI opens what is likely an even more vast universe of possibilities.

https://jared.xyz