Morgan Stanley points to four “emerging products” as Meta looks to convert AI spending into earnings

Meta Platforms is facing skepticism on Wall Street about whether its steep artificial intelligence spending will translate into near-term financial returns, but Morgan Stanley says the company’s competitive moats may be expanding faster than investors expect. In a research note cited by, analyst Brian Nowak maintained Meta’s “top pick” status and a $775 price target, arguing that four nascent product areas could each add $1 to $3 to Meta’s 2028 earnings per share and support valuation multiple expansion. The argument is that Meta’s AI buildout is not only about model training, but also about monetizing new user experiences and improving the efficiency of its core ad engine.

According to the same report, investor sentiment has lagged other megacap peers because of limited visibility into the return on an estimated $380 billion capital expenditure plan for 2027 and 2028. Meta’s stock, the article notes, has underperformed the S&P 500 over the past year, and a sharp decline on Friday was tied to a Financial Times report that Meta is considering a major equity raise to fund its AI infrastructure buildout. The takeaway for shareholders is that even optimistic long-range scenarios are colliding with short-term questions about how much dilution or financing risk may be involved.

The first “catalyst” Morgan Stanley highlighted is a multi-billion-dollar search opportunity integrated into Meta AI. The note’s modeling assumes the search feature could retain about 1 billion users and monetize roughly 10% of daily queries, which the article says could translate into more than $10 billion in annual revenue. Search is a particularly sensitive battleground because it can change how users discover content and products, and it can create new monetization surfaces beyond traditional feed-based recommendations and ad targeting.

Second, the bank’s framework focuses on higher-margin revenue streams through tiered subscription offerings. Morgan Stanley’s estimates in the cited write-up suggest that packaging subscriptions for Meta’s 3.5 billion daily active users could bring in about $7 billion of incremental revenue if adoption occurs among both its 50 million content creators and advertisers. The firm links these additions to incremental earnings per share potential, framing subscriptions as a way to diversify income and reduce reliance on pure advertising cycles.

Third, Morgan Stanley points to continued upgrades to Meta’s advertising engine driven by GPU-enabled machine learning changes. The report says every 1% upside to modeled 2028 ad revenue could contribute $3.5 billion to revenue and lift earnings per share by about 2.5%. Meta’s own engineering write-ups support the direction of travel: in 2026, Meta described scaling its ads recommendation runtimes toward LLM-scale complexity with latency-focused optimizations, including techniques to improve hardware utilization and keep inference within tight service budgets.

Meta also previously described Andromeda, a personalized ads retrieval engine designed to improve ad matching efficiency using NVIDIA Grace Hopper Superchip hardware and system-level optimizations, with claimed improvements in recall and ads quality on select segments. These kinds of changes matter because the ad market rewards better return on ad spend, and improvements in retrieval and ranking can increase advertiser outcomes without proportionally increasing costs, which is crucial when infrastructure spending rises. Morgan Stanley’s point is that ad efficiency gains could amplify the financial payoff from Meta’s AI investments.

Fourth, the note discusses what it calls a “Neocloud safety net,” a concept that would give Meta flexibility to lease unused AI compute capacity to external enterprise clients. The article’s modeling says leasing even 1 gigawatt of compute capacity at standard market rates could cushion free cash flow and provide a minimum 8% upside to 2028 earnings. It is important to treat this as an analyst scenario rather than a confirmed company plan, since the article describes the idea in Morgan Stanley’s framing and does not cite a specific Meta commitment.

Meta does have a disclosed pattern of spending toward generative AI and infrastructure, which makes the underlying premise plausible even if the monetization details remain uncertain. In its 2024 results reporting, Meta said 2025 capital expenditures were expected to be $60 billion to $65 billion, with growth driven by increased investment to support generative AI efforts and the core business. Still, the specific assumptions behind Morgan Stanley’s four emerging products, including pricing, adoption rates, monetization paths for Meta AI search, and any concrete terms for compute leasing, are not spelled out in Meta disclosures cited by this report. What to watch next is whether Meta provides clearer timelines and public metrics for these initiatives, and whether the company addresses financing questions raised by the equity raise speculation.