Kim Forrest

There is concern about whether chipmakers can meet demand, but the memory market is an oligopoly with few players. Whether the market demands "Training Chips" (Nvidia) or "Inference Chips" (for running models), *both* require massive amounts of memory. Demand is agnostic to *which* AI chip wins. High demand + limited competition (Moat) = Bullish for memory makers (implies MU / SK HYNIX). Cyclical downturn in consumer electronics dampening non-AI demand.

Memory has entered a period of persistent undersupply after its historical boom-bust cycles, and SanDisk is using contracts to smooth that cycle, reducing investor whipsaw and supporting the stock.

Apple's strategy of restraint in AI is a winning one: as a hardware provider, it does not need to build its own models but can let the end user choose any AI that works on their device. This positions Apple as the likely star of the earnings season, avoiding the capex risk that hurts other tech names.

Amazon is a holding in the firm’s portfolio and its AI spending is aimed at satisfying its customer base in a non-circular way. Unlike NVIDIA, Amazon's customers will pay for the services, making the spending recoverable and a reasonable growth investment.

Synopsys has been sold off as if no more chips will be designed, which is unlikely. The company is a bargain at current levels and is being added to accounts.

Поставки памяти в настоящее время являются «узким местом» для ИИ, и немедленного решения этой проблемы не существует, поэтому в краткосрочной перспективе Micron будет демонстрировать высокую маржинальность. Компания наращивает производственные мощности, что свидетельствует об уверенности в будущем спросе. В долгосрочной перспективе технологии могут предложить альтернативные решения, но этот риск не является угрозой в ближайшем будущем. Сосредоточьтесь на заказах и наращивании мощностей как на опережающем индикаторе.

Intel is being revived by the forward movement of AI development. After being left for dead, it is now coming back as AI rollout benefits more than just NVIDIA. We are in the bottom of the second inning of AI, implying a long runway. Discernment is happening and Intel is a beneficiary.

Software, especially enterprise software, is important and the recent devaluation is due to a misunderstanding of its value. AI will not replace the need for software that stores and accesses historical data (like Microsoft Excel). The sell-off driven by AI hype is premature.

Microsoft stock has been beaten up alongside other software names due to fears of AI replacing human workflows. Forrest argues MSFT holds the "data repository for most businesses around the world." Unlike generic software, MSFT's moat is the data itself. It will not get "vibe coded out of existence." The sell-off is an overreaction; MSFT is a value-add company on sale. Enterprise spending slowdown or faster-than-expected AI agent adoption bypassing SaaS UIs.