$SPACEX - SpaceX is positioned as a must-own bet on both space and AI futures, with multiple revenue drivers including Starlink, terrestrial AI compute (EWS), and the frontier XAI model enhanced by Cursor acquisition. The company is monetizing AI compute at best-in-class rates ($50B/gigawatt with Google vs industry $14B) and has rapidly become the #4 hyperscaler.
$ANTHROPIC - Anthropic is demonstrating exceptional revenue growth and model capabilities with Fable 5/Mythos, maintaining Pareto dominance in coding and capturing the vast majority of frontier model revenues. The company has more proprietary coding tokens than exist on the public internet and is achieving accidental profitability ahead of expectations.
$NVDA - NVIDIA has maintained and even gained market share against ASICs contrary to expectations, with customers emphasizing they're consuming more NVIDIA than anticipated. The company has the optionality to release frontier open source models and compete directly with customers if economics change, while maintaining superior tokens-per-watt economics.
$AICOMPUTE - AI compute demand is far outstripping supply with monetization rates per gigawatt increasing from $20B to $30-40B, while less than 0.2% of people are using AI agentically. Token pricing is inflating rather than deflating due to supply-demand dynamics, and long-running models are unlocking massive new token consumption.
$AILABS - Frontier AI labs are capturing 90%+ of economic value despite open source consuming more tokens, with revenue trajectory from $200B end of 2025 to potential trillions by 2029-2030. The math supports $1.5T in 2027 CapEx against $300B+ inference revenue at 60-70% gross margins.
$COREWEAVE - CoreWeave represents a major business in the neo cloud space, benefiting from the massive shortage in AI compute and the need for rapid deployment capabilities that hyperscalers cannot match.
Bearish:
$ASICS - Custom ASICs from Meta and Microsoft have been disappointing, with NVIDIA maintaining and gaining share contrary to expectations. The ASIC thesis of taking significant share from NVIDIA has been decisively wrong, with NVIDIA likely to maintain much higher than the 30% implied market share.