May 1, 2026, 6:08 PM ETInterview, Fireside Chat, Conference Presentation

OpenAI Misses Targets, Codex vs Claude, Elon vs Sam Trial, Big Hyperscaler Beats, Peptide Craze

OpenAI missed its 2025 targets for ChatGPT regarding both weekly active users (1 billion target) and revenue, with the 1 billion user milestone now four months overdue in 2026.
The company faces a structural imbalance where $600 billion in future compute spending commitments exceeds the current valuation of the enterprise on secondary markets.
A conflict exists between CFO Sarah Fryer, who believes the company is unready for public reporting standards, and CEO Sam Altman, who is pushing for an IPO later this year.
Polymarket odds for an OpenAI IPO by the end of 2026 have dropped to 32% from 60% in December, reflecting growing uncertainty about capital allocation.

Product sentiment is shifting in OpenAI's favor with the release of GPT-5.5, which is receiving strong developer feedback and is outperforming the newly released Anthropic Opus 4.7, which users report is rationing compute and reducing thinking time.
The primary constraint in the AI market is identified as access to power and electrical grid infrastructure, not demand, with less than half of announced power projects currently under construction due to supply chain delays and red tape.
OpenAI currently holds a compute advantage over Anthropic, which is described as "token constrained" and engaging in compute gating, whereas OpenAI has excess capacity from its data center investments.
Google Cloud revenue grew 63% year-over-year to $20 billion in a quarter, and Microsoft Cloud grew 30% to $34.7 billion, validating the massive capital expenditure (CapEx) strategy across hyperscalers.
The "Rule of Three" thesis suggests AI markets will evolve into a 4-2-1 split, with OpenAI and Google fighting for consumer dominance and Google/Anthropic leading the enterprise sector.
MIT research on neural network pruning indicates potential to reduce model size by 90% without accuracy loss, which could lower inference costs by 10x and mitigate energy constraints.

Elon Musk's lawsuit against OpenAI is a bench trial in Oakland seeking $150 billion in damages, alleging breach of charitable trust and unjust enrichment for converting a non-profit into a for-profit entity.
Discovery revealed a diary kept by former OpenAI employee Greg Brockman documenting plans to remove Musk and turn the entity into a for-profit without his consent, described by the hosts as a "smoking gun."
Polymarket odds currently sit at 42-43% that Musk will win the case, though speculation suggests a settlement where he receives a share of equity rather than the full $150 billion.
Reports indicate OpenAI previously offered Musk shares, which he declined because he wanted the entity to remain a charitable organization to save humanity, fearing Microsoft control in a for-profit structure.

OpenAI released GPT-5.5 Cyber, a model capable of completing multi-step cyber attack simulations, matching the capabilities of Anthropic's Mythos proof-of-concept model.
Hosts argue that AI cyber capabilities will not create new vulnerabilities but will discover dormant bugs already present in code, leading to a one-time massive upgrade cycle for global software security.
The cybersecurity market is expected to evolve from human-to-human attacks to machine-to-machine attacks, necessitating a rewrite of the world's operational software to become impregnable.
Palo Alto Networks and CrowdStrike are positioned to capitalize on this shift, with CrowdStrike reportedly stating they can penetrate and manipulate any existing AI model.

Eli Lilly's Phase 3 data for Retatrutide (RETA) shows an average weight loss of 37 pounds in 40 weeks, with significant reductions in liver fat (80%), non-HDL cholesterol (27%), and A1c levels.
Retatrutide binds to three receptors (GLP-1, GIP, and glucagon), distinguishing it by increasing metabolism to favor fat burning over muscle loss, potentially making it a "de-aging" drug.
Projected commercial availability for Retatrutide is mid-2027, with the drug positioned as a premium "upside" product alongside the lower-cost Zepbound.
Host Chamath Palihapitiya disclosed personal use of "Sparks," a supplement from Rho, describing it as requiring significant scheduling and preparation for optimal effect.

The Supreme Court heard oral arguments in a case involving Bayer/Monsanto and the EPA's authority to set pesticide labels, with the White House arguing for federal preemption against state "failure to warn" laws.
The case hinges on the legal interpretation of whether states can override EPA determinations on product safety labels following the overturning of the Chevron doctrine.
The court hearing is expected to be a coin-flip decision (5-4 or 4-5), with significant implications for the power of federal regulatory bodies like the EPA and FDA versus state authority.
The Supreme Court is currently hearing only two cases per day (60 minutes each) between October and April, with strict protocols regarding silence and conduct inside the marble building.

The four largest hyperscalers (Amazon, Microsoft, Google, Meta) announced a combined $725 billion in CapEx guidance for 2026, signaling a shift from asset-light software models to heavy infrastructure investment.
This massive investment is causing a dramatic drop in free cash flow, with Amazon's free cash flow down 97% and Google/Meta down double-digit percentages as they prioritize infrastructure build-out over shareholder buybacks.
Hosts note that unlike the dot-com bubble's "dark fiber" scenario, current AI CapEx is driven by immediate, voracious demand for compute and tokens, preventing the risk of unused infrastructure.
The AI sector is currently contributing 75% of US GDP growth in the quarter, linking the technology boom directly to the broader American economic expansion.

An incident at Pocket OS involved an AI agent deleting an entire production database and backups due to a credential mismatch and lack of human supervision, highlighting the risks of "vibe coding" without oversight.
The hosts discussed the "peak of inflated expectations" regarding AI agents, noting that while productivity will increase, full automation of jobs is unlikely due to the need for human accountability in complex workflows.
Hosts debated the economics of luxury travel, with Jason Calacanis spending $30,000 on courtside tickets and Chamath noting he would allocate funds to a hotel suite but acknowledging the event sold out all inventory due to demand.