AI Trade Shifts from GPU Focus to Sustainable ROIC

GPU shortages and supply bottlenecks initially anchored the AI trade, as Vince Lorusso notes that investors previously sought broad exposure to all semiconductor and memory firms.

Conversely, there was a widespread fear of disintermediation by AI, leading investors to short software-related entities, but Lorusso suggests the new thesis demands a focus on companies with sustainable competitive advantages.

Clough Capital's current focus is on navigating the long-running cyclicality in chips and memory by identifying investable companies that are positioned at critical infrastructure bottlenecks.

The aim is to find entities capable of delivering high returns on invested capital (ROIC) well into the future, moving away from broad market signals to a more detailed, bottom-up fundamental analysis.

The Mag-7 grouping is no longer considered a monolithic entity, as individual companies within it, like Meta, are demonstrating distinct business models and strategic advantages, known as 'moats'.

Meta, once primarily a social media company generating revenue from WhatsApp, Instagram, and Facebook, is evolving to become a foundational 'railroad' or infrastructure provider for AI, and the market is now discerning between software winners and losers, such as those in cybersecurity.

Individual company performance is diverging from broad market trends, leading Vince Lorusso to state that investors should adopt a more tactical stance during these middle innings of the cycle.

Software stocks have recently experienced struggles, trading 20% below their previous highs, indicating a broader challenge in the sector.

AI agents pose a significant risk to existing software services, particularly those with seat-based licensing models, as these agents could potentially replace services offered by companies like Adobe.

Despite initial broad sell-offs, cybersecurity remains a strong conviction area for investors because AI increases the number of endpoints and identities that require robust protection.

Investors are advised to prioritize ownership of the 'toll road' of AI infrastructure rather than entities involved in lending against it, suggesting a cautious approach to financing-related names such as Oracle, SoftBank, and Blue Owl.

Late-stage consumer companies are identified as another sector requiring careful monitoring due to potential headwinds and changing market dynamics.

While memory remains a cyclical industry, it could be experiencing a potentially longer, record-breaking cycle, with market attention focused on return on invested capital and the specific demand for high-bandwidth memory.

The recent appreciation in stock prices makes it more challenging to reconcile expectations for future demand growth, and Micron's upcoming earnings report will provide crucial data points to assess the broader AI trade's strength.