ChatGPT's Viral Explosion: 1 Million Users In... 60 Minutes

This episode dissects the explosive growth and market volatility at the intersection of AI and crypto, from OpenAI's viral user acquisition and massive valuation jump to CoreWeave's turbulent IPO signaling strong underlying demand for AI compute.

OpenAI's Viral Explosion and Valuation Surge

• The conversation kicks off discussing the viral "Giblification" trend, where users applied a Studio Ghibli-style artistic filter to images using a new OpenAI feature. Jaws notes this animation style, originating from the famed Japanese animation house Studio Ghibli, rapidly dominated social media feeds, even being applied to contentious events like political moments and historical tragedies, highlighting the unexpected power and reach of accessible AI tools.
• Behind the viral fun, significant business developments unfolded for OpenAI. Sam Altman highlighted the staggering acceleration in user acquisition: while ChatGPT took 5 days to reach 1 million users ~26 months ago, the new image animation feature hit 1 million users in just 60 minutes, two days post-launch. This rapid adoption coincided with OpenAI closing a new funding round, reportedly valuing the company at $300 billion, alongside a massive $40 billion capital raise – noted by CNBC as the largest private tech funding round ever, underscoring immense investor confidence despite broader market anxieties. David points out this signifies robust health in the private AI investment market, seemingly detached from recent public market downturns.

CoreWeave's Volatile IPO: A Litmus Test for AI Markets

• The discussion shifts to CoreWeave, a company with deep crypto roots, whose recent NASDAQ IPO served as a key indicator for AI market sentiment. Founded in 2017 as Atlantic Crypto for Ethereum GPU mining, CoreWeave pivoted during the 2018 crypto winter, leveraging its GPU infrastructure for general high-performance computing (HPC) tasks like graphics rendering before the AI boom fully materialized.
• Today, CoreWeave operates data centers across the US and Europe, providing Nvidia GPU-based cloud computing for AI model training and inference (the process of using a trained AI model to make predictions or generate outputs).
• Despite reporting $1.9 billion in 2024 revenue (with a $900 million net loss due to scaling investments) and raising $1.5 billion in the largest AI-related IPO to date, CoreWeave's debut was shaky. Opening below its $40 IPO price, commentators initially flagged concerns: high debt ($8B + $2.6B lease liabilities), heavy customer concentration (77% revenue from Microsoft and OpenAI), potential lack of a defensible business moat beyond "GPUs in the cloud," and volatile market timing. The Financial Times initially declared, "CoreWeave's IPO tested the waters and missed the mark." However, sentiment dramatically reversed on day two, with the stock ripping 42% higher, eventually surging from a low of $36 to over $65 before settling around $56, adding billions in market cap.
• Jaws and David interpret this volatile price action, resembling a crypto bull market chart, as a strong positive signal for underlying AI demand, potentially validating the immense need for inference compute power highlighted by OpenAI's "melting GPUs" during the Giblification craze.
• A fascinating historical footnote reveals CoreWeave's survival was partly enabled by delays in Ethereum's transition to Proof-of-Stake. The prolonged GPU mining profitability allowed CoreWeave to sustain operations and stockpile high-end Nvidia GPUs until the AI inference market, catalyzed by ChatGPT's 2022 launch, became viable. This blend of foresight (anticipating future compute demand) and fortune (Ethereum merge delays) was crucial to their success.

xAI Acquires X: Strategic Implications for Data and Distribution

• In a significant restructuring, Elon Musk's AI company, xAI (creator of the Grok AI model), has acquired X (formerly Twitter). While previously intertwined with xAI as a subsidiary, X is now owned by the AI lab. Musk framed this as combining "data, models, compute, distribution, and talent," valuing the combined entity with xAI at $80 billion and X at $33 billion in an all-stock transaction.
• Jaws emphasizes the strategic importance: xAI now has unparalleled, direct access to X's massive real-time data stream (600 million active users' tweets, opinions, and breaking news) and a vast distribution network. This positions Grok uniquely, potentially enabling superior Retrieval-Augmented Generation (RAG) – a technique where AI models retrieve external, up-to-date information to enhance their responses – making Grok highly informed on current events.
• David notes Twitter's role as the primary source for real-time information, suggesting Grok could leverage this unique, high-fidelity data moat. Furthermore, xAI announced Grok integration into Telegram (with its reported 1 billion monthly active users), signaling aggressive market penetration and potential for training on diverse, personalized chat data, further enhancing the model's capabilities and reach.

Google Gemini 2.5 Pro Takes the Crown (For Now)

• The weekly "best AI model" title shifts, with Google's Gemini 2.5 Pro emerging as a top contender, particularly praised for its coding capabilities, reportedly surpassing Anthropic's Claude, which previously held a strong position among developers. Key strengths highlighted include its powerful performance and a massive 1 million token context window (the amount of information, roughly 600k-700k words, the model can consider in a single prompt), enabling complex tasks and longer memory retention crucial for applications like personalized coaching or detailed analysis.
• A Twitter discussion highlighted a divergence: while OpenAI's Giblification captured public attention ("vibes"), developers and businesses ("people who pay with their credit cards") are noticing Gemini 2.5 Pro's strengths, particularly its cost-effectiveness and speed compared to competitors. This suggests a split between consumer-facing mindshare (where ChatGPT excels) and B2B/developer adoption (where Gemini is making inroads).
• Peter Yang's comparison chart visually reinforces this dynamic landscape, showing different models leading in specific areas (e.g., ChatGPT strong in image generation, Gemini in coding), while also revealing feature gaps across models (like deep research, video generation) that likely represent the next frontiers of development.

The Rise of Vibe Coding: Cursor's $9.6B Valuation

• The concept of "Vibe Coding" – using natural language prompts to generate code – gains significant traction with Cursor, the company behind it, achieving a near-$10 billion valuation ($9.6B) after a $625 million funding round led by prominent VCs like a16z.
• A demo showcased creating a 3D Flappy Bird clone using simple prompts and visual flowchart manipulation within Cursor, illustrating the dramatically lowered barrier to software creation.
• This trend echoes Andrej Karpathy's earlier observation that "English is the hottest programming language." David draws a parallel to the creation of the MOBA game genre (like Dota/League of Legends) from a Warcraft map editor, suggesting that simplifying creative expression tools can unlock massive unforeseen innovation. Jaws posits that AI might be accelerating the typical tech cycle, potentially flipping it so that consumer applications emerge rapidly and drive infrastructure demand, not just in software but also in media (e.g., AI-generated films), heralding a new "application era."

Crypto AI Market Update: Sentiment, Performance, and Key Trends

• Despite broader market volatility, the Crypto AI sector showed signs of life during a recent Bitcoin pump (briefly touching $88k). Data highlighted by analyst 0xJF indicated significant altcoin inflows disproportionately favored AI-related tokens, reinforcing the narrative that Crypto AI remains a dominant theme driven by market sentiment.
• Top performers included AI agent platforms (BID, Billy Bets), DeFi abstraction tools (NUR, Banker, Grift), robotics protocols (SAM), and crypto-native coding tools (Alchemy). The BankerBot interaction with Caitlyn Jenner served as a viral example of crypto AI accessibility.
• However, the subsequent Bitcoin retracement pulled many of these gains back. Key categories showing relative strength were identified as gambling-focused agents (like Billy Bets) and DeFi abstraction platforms. Conversely, categories like launchpads and major ecosystem/framework tokens (Virtuals, Arc AI) reportedly lagged in recovery.
• The pump in Faulcoin ($600M market cap) was noted, likely driven more by its meme status than pure AI fundamentals, highlighting the influence of memetics. The illiquidity of many smaller-cap AI tokens was also cited as a factor amplifying price swings, with Graham's tweet showing broad green performance in the Virtuals ecosystem during the pump potentially reflecting this.

Billy Bets Deep Dive: AI Agent vs. Human Gamblers in March Madness

• Ask Billy Bets, an AI gambling agent, gained prominence by competing in the "Profit X March Madness Tournament," pitting its AI-driven betting strategies against human sports bettors for a $25,000 prize. March Madness refers to the highly popular US college basketball tournament known for its unpredictable outcomes and widespread betting via "brackets."
• Billy Bets demonstrated strong performance, achieving a 7-3-1 record and advancing to the "Sweet 16" (final 16 competitors out of an initial 64 humans), generating roughly a 40% ROI on its $10,412 stake during the tournament period. This provides a compelling case study of an AI agent demonstrably competing, and partially succeeding, against humans in a complex, real-world prediction task.
• The team behind Billy Bets plans to scale by expanding into other sports leagues (football, hockey, etc.) and launching a "proxy betting" feature, allowing users to allocate capital for the agent to manage and bet on their behalf, potentially turning Billy Bets into a decentralized hedge fund for sports betting. Discussions also touched on integrating with traditional betting platforms like DraftKings, vastly expanding its operational reach and data inputs beyond its current reliance on on-chain prediction markets.

Olas Protocol: Demonstrating Decentralized Agent Growth

• Olas, described as an "OG" autonomous agent protocol contemporary to Bittensor, received attention for its focus on enabling users to build, own, and monetize AI agents via its open network. The ecosystem involves various stakeholders, including agent builders (Operators) and different types of agents (autonomous, semi-autonomous) performing tasks primarily in prediction markets and DeFi (e.g., their "Baby Degen" agent for cross-chain transactions).
• Olas stands out by providing a public dashboard showcasing tangible growth metrics for its decentralized agent network. Key figures include over 5.3 million cumulative autonomous agent transactions and 3.5 million mech requests (on-chain requests made by agents to access external AI tools or data via Olas's marketplace, paid for in crypto). This transparency addresses a common challenge in the space: quantifying actual agent activity and utility beyond simple token price or TVL metrics.
• While most active on Gnosis chain (a hub for prediction markets), Olas operates across nine chains. The team is reportedly working on V2 and refining tokenomics to better capture value for token holders, making it a project for researchers to monitor for developments in decentralized agent coordination and on-chain AI service economies.

Vanna Network: Training AI on User-Owned Private Data

• Vanna Network aims to tackle data ownership by creating a decentralized network where users can pool their private data (from sources like Reddit, location services, potentially chat logs) and get compensated when it's used for training AI models or sold. This contrasts sharply with traditional AI models trained primarily on publicly scraped internet data.
• Vanna announced "Collective One," their first AI model trained exclusively on this user-contributed private dataset, facilitated through a partnership with decentralized training network Flower Labs. While relatively small (7 billion parameters, suitable for mobile devices), the model's potential advantage lies in the unique nature of its training data, which could enable more specialized, personalized AI experiences currently difficult to achieve with generalist models.
• This initiative taps into the growing interest in decentralized AI training and alternative data sourcing. However, Jaws raised practical questions about the economic sustainability – whether the compensation offered to users for contributing highly personal data will be sufficient to incentivize participation at scale.

Conclusion

This episode highlights AI's pervasive influence, driving compute demand reflected in markets like CoreWeave's IPO, while specialized crypto AI applications like Billy Bets and Olas demonstrate niche traction. Investors and researchers should monitor both AI infrastructure plays and the evolving utility of decentralized agents and novel data models.