Morgan Stanley’s three preferred AI stocks

Morgan Stanley named Alphabet, Amazon and Meta as its three preferred large-cap AI stocks in a July strategy note, saying investors may rotate capital from semiconductor suppliers back to large cloud and internet companies.

The bank cited a June surge in semiconductor shares and renewed focus on the cost hyperscalers face when building AI infrastructure. The July 6 note said some money could move toward large cloud operators because those companies have underperformed recently and may show more discipline in capital spending in the near term.

Morgan Stanley internet analyst Brian Nowak outlined specific cases for each company. For Alphabet, Nowak highlighted the commercialisation of Google’s tensor processing units, or TPUs. Google built TPUs to run its own AI services and is offering the chips and related cloud capacity to third parties. In a June 29 note, Nowak raised his price target for Alphabet to $415 from $375 and estimated that sales of TPUs to outside customers could generate as much as $80 billion in revenue by 2028. Morgan Stanley warned Alphabet’s rising infrastructure spending could pressure cash flow until new capacity is monetised.

Amazon’s profile rests on Amazon Web Services and the company’s retail ecosystem. Morgan Stanley said AWS could meet stronger demand as Amazon brings additional data-centre capacity online and serves AI workloads that were previously constrained by capacity limits. Nowak kept an Overweight rating on Amazon with a $300 price target and modelled the potential for AWS growth to accelerate beyond 30% if returns on infrastructure investments improve modestly.

Meta was described as the most direct recovery candidate among the three. Morgan Stanley maintained a $775 price target while noting uncertainty about the timing of returns from the company’s large AI spending. The bank said Meta’s advertising business already benefits from AI through improved content recommendations and ad targeting on Facebook and Instagram. Morgan Stanley listed possible new revenue sources including Meta AI search, paid subscriptions, business-focused agents and the sale of excess computing capacity. Nowak estimated that four emerging products could each add between $1 and $3 to Meta’s 2028 earnings per share if adoption meets expectations, while flagging execution risk until revenue from these initiatives appears.

The July strategy note did not present the three stocks as a formal investment basket. Morgan Stanley framed the discussion as a response to changing sector performance: stronger semiconductor returns in June and recent underperformance among hyperscalers, along with potential stabilisation in hyperscalers’ capital spending and ways to monetise AI assets.