Investors should prioritize Anthropic (private) as it captures the enterprise market with a massive revenue ramp and a potential $600 billion IPO on the horizon. For accredited investors, the current "vibe shift" and negative sentiment surrounding OpenAI (private) creates a strategic entry point in secondary markets ahead of their next major model release. NVIDIA (NVDA) remains the essential "infrastructure play" as the transition to Blackwell architecture fuels the next exponential growth phase for all major AI labs through 2025. In the public markets, look for AI-driven cybersecurity firms and distributed AI projects like BitTensor (TAO) that offer unique protection against emerging digital vulnerabilities. While traditional software stocks in the IGV ETF face pressure, established players like Oracle (ORCL) and Salesforce (CRM) may offer value if they successfully pivot to "agentic" AI task automation.

Anthropic (Private)

Anthropic is currently experiencing what the analysts describe as the "fastest revenue explosion in the history of technology." The company has shifted its focus heavily toward enterprise coding and "co-work" tools, moving away from multimodal experiments like video to dominate the developer market.

• Revenue Ramp: The company’s revenue run rate reportedly topped $30 billion as of March 2024, up from $1 billion at the end of 2023.
• Enterprise Adoption: Over 1,000 enterprises are currently paying more than $1 million annually for Anthropic’s services.
• Market Share: Analysts estimate Anthropic currently provides 50% to 60% of all AI coding tokens in the market.
• Mythos Model: Anthropic is sandboxing its newest model, Mythos, for 100 days. The model autonomously found vulnerabilities in major operating systems (Linux, OpenBSD) and web browsers that had been missed for decades.
• Strategic Pivot: By focusing strictly on text and code rather than video or hardware, they have achieved a "step function" improvement in intelligence that is driving massive corporate demand.

Takeaways

• Enterprise over Consumer: While OpenAI (ChatGPT) has the brand recognition, Anthropic is winning the "wallet share" of the Global 2000 by focusing on high-utility labor augmentation.
• IPO Watch: Prediction markets currently give Anthropic a 51% chance of IPOing before 2027, with a potential valuation exceeding $600 billion.
• Efficiency Play: With only 2,500 employees, the revenue-per-employee ratio is unprecedented, suggesting high future profitability once compute costs stabilize.

OpenAI (Private)

Despite recent "FUD" (Fear, Uncertainty, Doubt) and high-profile departures, OpenAI remains a primary pillar of the frontier model space with a massive user base and upcoming high-end models.

• New Model "Spud": OpenAI is reportedly preparing to release a new model (internally referred to as Spud), trained on NVIDIA’s Blackwell chips. It is expected to be on par with Anthropic’s Mythos.
• Market Sentiment: The company is currently seeing some "hand-wringing" from employees and secondary market investors, with shares occasionally trading lower than previous valuations due to the aggressive rise of Anthropic.
• Codex Momentum: Their coding tool, Codex, continues to see an "incredible ramp" in usage despite the competition.

Takeaways

• Don't Count Them Out: Analysts suggest we are at "peak OpenAI FUD," and the release of their next frontier model could quickly shift the narrative back in their favor.
• Secondary Market Opportunity: For accredited investors, the current "vibe shift" might represent a more attractive entry point in secondary markets compared to the peak hype of 2023.

NVIDIA (NVDA)

The "chip layer" remains the fundamental backbone of the AI trade, with the transition to new architecture driving the next wave of model capabilities.

• Blackwell Architecture: The next generation of frontier models (like OpenAI's Spud) are being built on Blackwell, which is expected to unlock further exponential growth in model intelligence.
• Compute Constraints: Both Anthropic and OpenAI are currently "compute constrained," meaning they could grow even faster if they had more access to NVIDIA hardware.

Takeaways

• Infrastructure Supercycle: The transition from 1.5 gigawatts to 3+ gigawatts of compute per major lab suggests that the demand for high-end AI chips will remain robust through 2025.

Investment Themes & Sectors

AI-Driven Cybersecurity

The discussion highlighted a "Code Red" moment for global digital infrastructure.

• The Insight: AI models are now as good as professional humans at identifying bugs. We are entering a "pre-release" window where companies must use AI to patch 20-year-old vulnerabilities before hackers use the same models to exploit them.
• Actionable: Look for cybersecurity firms that are integrating "AI-driven defense" to counter "AI-driven offense."

The "Agentic" Era

The focus is shifting from AI that answers questions to AI "agents" that perform tasks.

• The Insight: Coding is the "step up" to agents. Because agents need to write code to execute tasks, the leaders in coding models (Anthropic, OpenAI) are best positioned to dominate the multi-trillion dollar agent market.
• Risk Factor: The "OpenClaw" controversy suggests that frontier model companies may "ankle" (limit) open-source competitors to protect their own agent ecosystems.

Open Source vs. Frontier Models

A major debate persists on whether "Frontier" (paid/closed) or "Open Source" (free/distributed) will win.

• The Insight: While open source is great for 70% of token consumption, large enterprises with "skin in the game" are unlikely to outsource their production code to open-source projects due to security and liability.
• Crypto-AI Convergence: Projects like BitTensor (TAO) and Venice are mentioned as "orthogonal attack vectors" that use distributed computing to challenge the $20 billion training costs of the big labs.

Enterprise Software (SaaS) Under Pressure

• The Insight: The IGV (Software ETF) has seen significant drawdowns. Investors are terrified that frontier models will "eat" the functionality of niche SaaS companies.
• Actionable: There may be "value buys" in enterprise software (e.g., Salesforce, HubSpot, Oracle) if they can successfully integrate AI, but the market is currently skeptical of their "moats."