As Meta's stock plunges, UBS and Deutsche Bank come out in support: AI capex is entering the harvest period, and five AI monetization paths have been launched
UBS raised its Meta target price from $884 to $892, and Deutsche Bank from $820 to $840, both maintaining Buy ratings, believing that Meta's AI capex over the past three years is shifting from cost pressure to revenue-generating assets. Meta's Family of Apps advertising revenue in Q3 was $63.1 billion, up 26% year-over-year, and its Q4 revenue guidance of $73 billion-$76 billion is above market expectations. About six weeks after launch, Muse has surpassed 10 million users in the US and Canada, and about 25% of users who have set it up use it on any given day.
Body · AI translation
Amid the plunge, Wall Street chose to stand behind Meta, betting that AI capital expenditure is entering a harvest period. UBS and Deutsche Bank both maintained Buy ratings and raised their target prices in their latest research reports. The core judgment is that Meta's AI capital expenditure over the past three years is beginning to enter a harvest period, with the advertising system benefiting first, and five AI commercialization paths already launched.
Overnight, the Nasdaq plunged 2.6%, and the Philadelphia Semiconductor Index fell 6.7%, putting pressure on the AI supply chain. Although Meta's third-quarter revenue and fourth-quarter guidance both beat expectations, its AI funding commitment surged to $825 billion, causing free cash flow to turn negative for the first time in fourteen years. The market is highly concerned that massive capital expenditure will bring short-term bond issuance pressure, and its five-year CDS soared to a record high. Dragged down by this, Meta's stock price fell about 7% on Thursday, with market value evaporating by more than $100 billion.
According to Chase Trading Desk, UBS Securities LLC analyst Stephen Ju raised Meta's target price from $884 to $892 in his latest report, maintaining a Buy rating. Deutsche Bank analyst Benjamin Black also raised the target price from $820 to $840. The common judgment of the two research reports is that the market's pricing logic for Meta's AI spending is changing, and computing power is beginning to shift from a cost pressure to a revenue asset.
For investors, the real watershed is the "speed of realization." Advertising remains Meta's hardest support, and the first stop for AI revenue is still the advertising system; if revenue brought by Muse, Business AI, search ads, Advantage+, and creator chatbots grows faster than depreciation and expenses rise, AI capital expenditure will be repriced. If revenue realization slows, current valuations will face heavier pressure.
Advertising remains the hardest support
Meta's Family of Apps advertising revenue in the third quarter was $63.1 billion, up 26% year over year, accounting for the vast majority of total revenue of $64.6 billion. Advertising remains the core foundation supporting valuation.
Breaking it down, ad impressions grew 10% year over year, lower than 18% in 2Q26 and 14% in 3Q25, indicating a marginal cooling in volume growth. But the average price per ad rose 14% year over year, offsetting the pressure from slowing impressions. Rising advertising spending for the U.S. midterm elections also supported ad prices.
Fourth-quarter guidance further reinforces revenue resilience. Meta expects revenue of $73 billion-$76 billion, with a midpoint of $74.5 billion, above the market expectation of $73.6 billion, and the guidance already includes about 1 percentage point of currency headwind.
Product-level AI improvements have begun to feed into advertising performance. After the Instagram Reels relevance upgrade, user time spent increased by more than 4%; Facebook's new generative retrieval architecture increased Reels watch time by 1%; Meta Generative Recommender increased conversion rates by nearly 2% in some targeted scenarios in Facebook Feed; the GEM expansion brought a 1.7% increase in off-site conversions and a 9.6% increase in app ad conversions; Andromeda improvements increased Facebook conversion rates by 1.4%.
There is also feedback from advertisers. Advertisers that adopted suggestions from Meta AI Business Assistant saw conversion volume increase by nearly 5% at the same cost per conversion, and their ad spend was 9% higher than comparable non-adopters. UBS noted that the incremental budget heat brought by Advantage+ is the highest of the year.
The change in Meta's valuation narrative lies in the fact that AI infrastructure has already pressed down on advertiser delivery performance, user time spent, and conversion rates. The core advertising system is already absorbing the dividends brought by computing power and model upgrades.
Deutsche Bank's Benjamin Black's judgment is that Meta is undergoing a turning point: "Computing power will shift from being viewed as a 'cost center' to being viewed as a 'revenue-driving asset.'" This sentence summarizes the current bull case: if AI infrastructure simultaneously improves ad conversion, extends user dwell time, and supports consumer-side agents and enterprise-side products, the market's discount on capital expenditure will decline.
Muse is one of the key variables. About six weeks after launch, Muse has more than 10 million users in the United States and Canada; among users who have set up Muse, about 25% will use it on any given day. Shopping has become one of the most common scenarios, with more than a quarter of users participating in shopping, and the average purchase amount exceeding $200.
This means Muse's value comes not only from chat usage but also from commercial intent. Meta has the opportunity to intervene in recommendation, comparison, and transaction links after users express purchase intent, and then feed high-intent data back into the advertising system to improve the efficiency of the existing advertising machine.
The first is the core advertising system. Muse is used for ad targeting and content recommendation, directly affecting the two key links of "to whom content is shown" and "to whom ads are sold." Advantage+ uses generative AI for creative and delivery automation to improve advertisers' budget efficiency.
The second is Business AI and Business Agents. They are tied to messaging ads, with the goal of increasing usage and monetization of WhatsApp and Messenger. Meta's Family of Apps has 3.64 billion daily active users, and at least 500 million entities self-identified as small businesses use these apps to reach customers and generate revenue. If consumer-side agents interact directly with merchant-side agents, Meta can put discovery, lead generation, transactions, business messaging, ad delivery, and merchant productivity tools into the same network.
The third is search and shopping ads. If Meta AI/Muse can capture purchase intent, it can move into a position closer to transaction decisions. Shopping and travel are already leading use cases for Muse, and monetization paths include transaction commissions, subscriptions, more advanced models or greater token usage, paid features for business owners, enterprise APIs, and Muse Code.
The fourth is creator chatbots. Chatbots aimed at influencers and creators can lengthen user dwell time and bring more monetizable surfaces. The logic is similar to Reels: first boost engagement, then gradually raise ad load and monetization efficiency.
The fifth is the enterprise side, APIs, and compute exports. Enterprise products in the near term include model APIs, Muse Code, and Meta Business Agents; the longer-term path is to deploy Muse into enterprise environments and expand Business Agents into functions such as sales and customer service. If the first few revenue lines materialize slowly, Meta can also sell surplus compute or model access rights in a supply-constrained environment.
Speed of realization becomes the dividing line
UBS did not fully write the optimistic AI scenario into its model. The latest estimates show Meta revenue is expected to be $256.2 billion in 2026, $317.3 billion in 2027, $379.2 billion in 2028, $440.6 billion in 2029, and $503.7 billion in 2030.
The upward revisions are relatively restrained. The 2027 advertising revenue forecast was raised from $304.5 billion to $308.6 billion, an increase of 1.3%; 2028 was raised from $361.3 billion to $367.1 billion, an increase of 1.6%. At the total revenue level, 2027 was raised from $314.3 billion to $317.3 billion, and 2028 from $375.6 billion to $379.2 billion, both increases of about 1%.
EPS changes are similarly limited. In the GAAP diluted EPS forecast, 2026 was cut from $29.28 to $29.04, 2027 was raised from $30.81 to $32.01, and 2028 was slightly raised from $31.08 to $31.11. The revenue upgrade is mainly reflected in 2027, while the EPS increase in 2028 is not obvious.
This shows that what the market is really focused on is the "speed of realization." Meta management hopes the cycle from product development to revenue contribution for Muse will be faster than the past product cycle of about 4 years. If AI revenue runs ahead of rising depreciation and expenses, capital expenditure can be regarded as upfront investment; if the pace of realization is slow, margins and free cash flow will continue to come under pressure.
The cost curve remains heavy, and 2027 is the key year
Meta narrowed its 2026 capital expenditure guidance to $135 billion-$145 billion and raised total expense guidance to $175 billion-$179 billion, including the previously disclosed $9.9 billion legal fine. Excluding that expense, the underlying expense guidance is unchanged.
In UBS's model, capital expenditure reaches $251 billion and $299.6 billion in 2027 and 2028 respectively, and the free cash flow yield on equity turns negative. Total expenses are forecast at $175.4 billion in 2026, $213.1 billion in 2027, and $266.2 billion in 2028. Depreciation and amortization also rises rapidly, to $26.6 billion in 2026, $45.8 billion in 2027, and $68.8 billion in 2028.
Deutsche Bank estimates show operating expenses of about $212 billion in 2027, up 20% year on year and above the market consensus of $204 billion; capital expenditure of about $233 billion, up 66% year on year and also clearly above consensus. Reality Labs is expected to post a full-year loss of $18 billion to $19 billion.
Efficiency improvements are key to buffering cost pressure. Management expects the computational cost efficiency of Muse to improve by more than 10 times over the next year, through levers including model distillation, higher GPU utilization, inference optimization, use of off-peak or idle compute, product-level efficiency improvements, and optimizing specialized models for specific tasks. Internal efficiency is also changing: software output per engineer has more than doubled year on year, and more than 70% of measurable analysis output is AI-assisted.
UBS raised its Meta target price from $884 to $892, based on a 29x price-to-earnings ratio applied to full-year GAAP diluted EPS of $30.77 through 3Q28. Based on the after-hours price of $712.29 on October 21, that target price implies about 25% upside.
A 29x valuation is not low. UBS's framework shows that this multiple is about 2 standard deviations above Meta's long-term average GAAP diluted EPS trading level. The premise supporting this valuation is that Business Agents are already widely available, Muse is already live with growing usage, and AI spending is showing signs of monetization.
UBS's bull case gives a target price of $1,206, implying about 69% upside. That scenario assumes two-year revenue compound growth accelerates to 30%, an operating margin of 34% in 4Q27-3Q28, EPS of $40.20, and a 30x PE.
The bear case is $539, implying about 24% downside. That scenario assumes two-year revenue compound growth slows to 15%, an operating margin of 25%, EPS of $21.56, and a 25x PE. The main pressures come from equity financing concerns after the introduction of Class C shares, the still-unclear framework for operating expenses and capital expenditure after 2027, and slowing growth in ad impressions.
For investors, the core of what Meta is trading on is now clear: advertising remains the hardest support, AI spending is being repriced as a revenue asset, and five AI revenue lines are starting up. What will determine the direction of the valuation next is whether these GPUs can generate sufficiently fast revenue returns across advertising, agents, messaging, search, and the enterprise side at the same time.
~~~~~~~~~~~~~~~~~~~~~~~~
The above excellent content comes from Chase Trading Desk.
For more detailed interpretation, including real-time commentary and frontline research, please join [Chase Trading Desk▪Annual Membership] Risk Disclosure and Disclaimer
Markets carry risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investment based on this is at the user's own risk.