Meta isn’t just spending billions on AI—It’s building multiple ways to profit from it

• Meta has underperformed the broader market in 2026 as investors remain sceptical that the company can generate sufficient returns on its aggressive AI investments.

• Meta is exploring two cloud service offerings—leasing AI computing capacity and offering hosted access to its AI models—which could create meaningful new revenue streams. We believe the company is more likely to follow the neocloud model than become a full-fledged hyperscale cloud provider.

• Meta has also expanded its AI monetisation strategy through subscriptions, with  Facebook Plus, Instagram Plus, WhatsApp Plus and Meta One plans that provide premium features and greater access to its latest generative AI models.

• The company is already delivering tangible returns through its advertising business, driving stronger user engagement, improved ad conversion rates and higher advertising prices, even before meaningful cloud or subscription revenue materialises.

• We maintain our target price of USD 1,018, implying 61% upside as of 9 July 2026, as we believe the market is overly focused on near-term AI spending while underappreciating Meta's growing AI monetisation avenues and long-term earnings potential.

Meta's shares are down 4.3% year-to-date, significantly underperforming both the S&P 500 (+10.2%) and the Nasdaq-100 (+17.7%), as investors remain unconvinced that the company can generate attractive returns on its aggressive AI investments. During its first-quarter earnings call, Meta raised its 2026 capital expenditure guidance from USD 115–135 billion to USD 125–145 billion, reiterating its commitment to expanding AI infrastructure.

However, since our last update, Meta has made meaningful progress in monetising its sizeable AI investments beyond its highly profitable advertising business. The company has launched new subscription offerings and is planning to commercialise its AI infrastructure via a cloud business.

Figure 1: Meta has underperformed year-to-date

Meta explores cloud services to monetise its AI infrastructure

On 1 July, Bloomberg reported that Meta was developing a cloud infrastructure business. The first offering would allow developers to access Meta's AI models through a cloud service, with Meta hosting the models on its own infrastructure and charging developers based on usage. This API strategy has since been confirmed with the launch of Muse Spark 1.1, Meta's latest AI model, for which the company said developers will be charged once usage exceeds a specified token threshold. The second offering reportedly under consideration is the leasing of raw AI computing capacity to developers, similar to specialist AI cloud providers such as CoreWeave and Nebius.

Previously, Meta had declined requests from external companies seeking access to its compute infrastructure, preferring to reserve capacity for internal AI development. The shift in strategy likely reflects management's desire to generate earlier returns on its sizeable AI investments. As investors continue to question how Meta intends to monetise tens of billions of dollars in AI capital expenditure, leasing AI infrastructure could provide a meaningful near-term revenue stream at a time when demand for AI compute remains exceptionally strong amid persistent shortages of power and advanced AI chips.

Importantly, we do not believe this necessarily suggests that Meta has built excess AI capacity. Instead, it may simply reflect a more efficient optimisation of its infrastructure. A useful comparison is SpaceX, which leased capacity at its Colossus 1 data centre after technical constraints—including latency issues and a mix of different Nvidia GPU generations—limited its internal utilisation. By renting the facility to Anthropic, SpaceX was able to monetise underutilised infrastructure while preserving newer, more strategically valuable capacity for its own AI development. Meta's situation could be similar. Indeed, a Meta spokesperson said on 7 July that the company is "still hungry for even more computing power." This was followed by Meta's announcement on 9 July that it plans to build its 33rd data centre in Canada, reinforcing the view that internal demand for AI infrastructure remains robust.

At the same time, we do not view Meta's plans to lease AI computing capacity as a sign that the company is stepping away from the frontier AI race. The company recently unveiled Muse Image, its latest image-generation model, which will be available to consumers through the Meta AI app and website, WhatsApp, and Instagram Stories. According to Meta's internal benchmark tests, Muse Image trails OpenAI's latest GPT Image 2 model but outperforms Nano Banana 2 in image editing tasks involving both single and multiple images. Meanwhile, Meta's upcoming Muse Video model ranked third on the Artificial Analysis Video Arena leaderboard, behind Google's Gemini Omni Flash and ByteDance's SeedDance, but ahead of Alibaba's HappyHorse. Muse Spark 1.1 also outperformed Google's Gemini across several benchmarks covering agentic capabilities, coding and multimodal reasoning. These developments suggest Meta remains firmly committed to improving the competitiveness of its frontier AI models rather than merely becoming an infrastructure provider.

In our view, selling AI compute is likely to remain a complementary revenue stream rather than develop into a full-fledged hyperscale cloud business. Competing with established cloud providers requires far more than building data centres. It also involves developing enterprise software platforms, developer tools, dedicated sales organisations and customer support capabilities. If Meta genuinely intended to compete head-on with hyperscalers such as AWS, Microsoft Azure and Google Cloud, we would expect to see substantially greater investment in these enterprise capabilities. Instead, the company has continued to streamline parts of its workforce while prioritising AI research and consumer-facing AI products. We therefore believe Meta is more likely to follow the model of neocloud players such as CoreWeave and Nebius, selectively monetising spare compute capacity while remaining focused on its longer-term objective of developing industry-leading AI models.

Subscriptions add another layer of AI monetisation

Beyond its planned cloud offerings, Meta has also begun rolling out paid subscription services across its ecosystem. In May, the company introduced Facebook Plus, Instagram Plus and WhatsApp Plus, alongside two subscription tiers for its Meta AI app and website.

Facebook Plus and Instagram Plus, priced at USD 3.99 per month, unlock premium features such as enhanced analytics, story rewatch statistics, greater audience reach and profile customisation tools aimed at creators and power users. Meanwhile, WhatsApp Plus, priced at USD 2.99 per month, focuses on personalisation features including premium stickers, custom ringtones and app themes.

More importantly, Meta has begun testing two subscription plans for its Meta AI app and website—Meta One Plus (USD 7.99 per month) and Meta One Premium (USD 19.99 per month)—in Singapore, Guatemala and Bolivia. The higher-tier plan provides users with additional computing capacity to generate more comprehensive AI responses and access advanced capabilities. Both Meta One plans also unlock greater access to the newly launched Muse Image and upcoming Muse Video models, allowing users to generate and edit more AI images and videos beyond the free usage limits. Both tiers also bundle Facebook Plus, Instagram Plus and WhatsApp Plus at a discounted price compared to subscribing to each service separately.

Crucially, these new subscription offerings build on Meta's already successful AI monetisation efforts within its advertising business. AI-powered recommendation and advertising models have increased user engagement, improved conversion rates for advertisers and strengthened pricing power, demonstrating that Meta's AI investments are already generating attractive economic returns even before any meaningful cloud or subscription revenue emerges.

During the first quarter, AI-driven content recommendations contributed to a 10% increase in Reels time spent on Instagram and more than an 8% increase in video time on Facebook. Stronger engagement enabled Meta to increase ad impressions by 19% year over year (YoY). At the same time, continued improvements to its AI-powered advertising systems increased conversion rates for landing page view ads by more than 6%, enhancing advertisers' return on ad spend. As campaign effectiveness improved, advertisers were willing to pay more for Meta's advertising inventory, driving a 12% YoY increase in the average price per advertisement.

Near-term headwinds create a compelling valuation opportunity

While Meta's expanding AI monetisation initiatives are encouraging, we expect investor sentiment to remain cautious in the second half of 2026 as the company continues to invest aggressively in AI infrastructure. Capital expenditure is likely to remain elevated over the coming quarters, weighing on free cash flow (FCF) and keeping investor concerns over the returns on AI spending firmly in focus. In fact, we believe Meta could report negative FCF in both 2026 and 2027. However, we expect FCF generation to improve gradually as capital expenditure growth eventually moderates—our base case is from late 2027 onwards—and the company's expanding AI monetisation initiatives, including advertising, subscriptions and potentially cloud services, begin to make a more meaningful contribution to earnings.

Despite the near-term pressure on cash flows, Meta remains well positioned to fund this investment cycle, supported by a robust balance sheet with a net debt-to-EBITDA ratio close to zero.

Related article: Digital Economy (Internet) Outlook 2H26: Better Late Than Never; Maintained 3.5 Stars

For now, we are maintaining our earnings forecasts as we await greater clarity from Meta's second-quarter earnings results in late July, particularly on the progress of its cloud strategy, the traction of its newly launched subscription offerings and their financial contribution. We see scope for upward revisions should these initiatives gain stronger-than-expected traction.

We therefore maintain our target price of USD 1,018, representing 61% upside from Meta's closing price of USD 631.48 on 9 July. In our view, the market remains overly focused on the near-term drag from AI capital expenditure while underappreciating the multiple monetisation avenues that Meta is steadily building. This disconnect has left the shares trading at an attractive valuation, creating an opportunity for long-term investors to accumulate the stock before the full earnings potential of its AI investments becomes more widely recognised.

Table 1: Projections for Meta’s earnings

Meta Platforms	2025	2026E	2027E	2028E
Earnings Per Share (EPS)	29.3	33.6	36.5	44.3
Earnings Growth YoY	22.3%	14.5%	8.6%	21.4%
PE Ratio (X)	22.5	18.8	17.3	14.3
Target Price (based on a fair PE of 23X)				1018
Upside Potential				61.3%
Source: Bloomberg Finance L.P., iFAST Compilations. Data as of 9 July 2026

Figure 2: Share prices are driven by earnings growth in the long run

Declaration:

This research report was prepared with the assistance of artificial intelligence (AI) tools. iFAST Financial Pte Ltd does not rely exclusively on AI for content generation; the content of this report – including all investment theses, ratings, price targets and conclusions – has been independently reviewed and verified by the research analyst(s) to ensure accuracy and professional integrity.

For specific disclosure, at the time of publication of this report, IFPL (via its connected and associated entities) holds a NIL position in the abovementioned securities. The analyst who produced this report holds a position in Meta Platforms.