Meta Platforms stands at a critical juncture, having reported a milestone first quarter in 2026 but with strategic tensions simmering beneath the surface. Just yesterday, on July 27, 2026, CEO Mark Zuckerberg's comments in the Financial Times, widely reported by Reuters, reiterated his position that the U.S. government "should not block Chinese models to gain an edge in the AI race." These remarks came alongside a Wall Street Journal op-ed where Zuckerberg framed the core AI policy question as one of "access and centralization," arguing for tools that "empower everyone" rather than being "restricted to a few institutions."
These statements land against the backdrop of Meta's massive and escalating investment in artificial intelligence. Its first-quarter 2026 results, announced on April 29, 2026, showcased robust financial performance, with revenue surging 33% year-over-year to $56.311 billion and diluted earnings per share reaching $10.44. However, the accompanying guidance dramatically raised Meta's capital expenditure outlook for 2026 to between $125 billion and $145 billion, a substantial increase from its prior range. This steep ramp-up in spending, primarily for AI infrastructure, is a direct reflection of Meta's ambition to "deliver personal superintelligence to billions of people," as Zuckerberg stated on the Q1 earnings call.
The convergence of Zuckerberg's strong public advocacy for open-access AI, his opposition to protectionist measures against Chinese models, and Meta's colossal AI infrastructure build-out creates a complex narrative for investors. The core question is how this open-access philosophy—which has famously driven the adoption of Meta's Llama models—will translate into sustainable competitive advantage and financial returns, especially as the company pours an unprecedented amount of capital into its AI future.
Meta’s financial performance in the first quarter of 2026 demonstrated strong top-line growth and impressive profitability. Revenue climbed 33% year-over-year to $56.311 billion, contributing to an operating income of $22.872 billion and a healthy operating margin of 41%. This growth was primarily fueled by the Family of Apps (FoA) segment, which generated $55.909 billion in revenue, dwarfing Reality Labs' (RL) $402 million. However, RL continued to be a significant drag on profitability, posting an operating loss of $4.028 billion in the quarter.
The most striking financial development is Meta’s escalating capital expenditure (CapEx). From $39.23 billion in 2024, CapEx surged 84% to $72.22 billion in 2025. The 2026 guidance, raised to $125-$145 billion, implies a further 73% to 101% increase year-over-year, with the midpoint of $135 billion representing an approximately 87% jump from 2025. This massive spending is already impacting cash flow: in Q1 2026, CapEx of $19.84 billion notably exceeded the free cash flow of $12.386 billion, signaling the intense front-loading of AI investments.
The following table summarizes Meta’s recent financial performance:
| Period (3M Ended) | Revenue ($M) | YoY Growth | Operating Income ($M) | Operating Margin | Diluted EPS | Capex ($M) |
|---|---|---|---|---|---|---|
| Q1 2026 (Mar 31) | 56,311 | +33% | 22,872 | 41% | 10.44 | 19,840 |
| Q4 2025 (Dec 31) | 59,893 | +24% | 24,745 | 41% | 8.88 | 22,140 |
| Q1 2025 (Mar 31) | 42,314 | +33% | 17,555 | 41% | 6.43 | 13,690 |
| Full Year 2025 | 200,966 | +22% | 83,276 | 41% | 23.49 | 72,215 |
These figures underscore a company performing strongly at the operating level, yet simultaneously embarking on an unprecedented CapEx journey that is already consuming more cash than it generates quarter-to-quarter. This trade-off between current profitability and future-oriented spending is at the heart of the investment thesis.
Meta's strategy for AI leadership has long centered on its commitment to open-access models, most notably the Llama family. This approach successfully fostered a massive ecosystem, with Llama and its derivatives reaching over 650 million cumulative downloads by December 2024, and exceeding one billion downloads by March 18, 2025. This rapid adoption, averaging roughly one million downloads per day since its initial release in February 2023, established Llama as a de facto standard in the open-weight AI community, generating tens of thousands of derivative models. As CEO Mark Zuckerberg noted in a 2024 blog post, "developers can run inference on Llama 3.1 405B on their own infra at roughly 50% the cost of using closed models like GPT-4o."
However, Llama's crown as the undisputed leader in open-weight models is now slipping. By September 2025, Alibaba's Qwen models had already surpassed Llama in cumulative downloads on Hugging Face, a key platform for open-source AI. This lead widened dramatically by March 2026, when Qwen tallied 942.1 million downloads compared to Llama's 476.0 million on the same platform, implying Qwen nearly doubled Llama's adoption in that six-month period. This shift is significant; Chinese models, including Qwen, reportedly captured over 50% of global open-source downloads by late 2025, with Chinese developers accounting for 63% of new derivative models.
Further complicating Meta's open-access narrative are the licensing terms for Llama 3 and Llama 4. These licenses include a critical restriction: a 700 million monthly active user (MAU) threshold, above which commercial users require a bespoke agreement. This effectively limits adoption by mega-scale platforms without direct negotiation, undercutting the perceived "openness" that developers might expect. This legal nuance, coupled with the rising prominence of more permissively licensed Chinese models, risks eroding Llama’s community appeal and limiting its deepest ecosystem penetration, even as Meta continues to laud its "open and decentralised innovation."
Meta's aggressive capital spending is not just a modest increase; it represents a full-throttle sprint in the global AI infrastructure arms race. The company’s CapEx trajectory has steepened dramatically, from $39.23 billion in 2024 to $72.22 billion in 2025, marking an 84% year-over-year increase. For 2026, the guidance of $125 to $145 billion implies a midpoint of $135 billion, which would be an astounding 87% jump from 2025 actuals. This level of investment positions Meta’s CapEx broadly in line with or ahead of some hyperscale cloud providers like Microsoft, despite Meta not operating a general-purpose public cloud business of comparable scale.
This enormous investment is explicitly directed towards building Meta's AI compute and data center capacity. As CFO Susan Li explained during the Q1 2026 earnings call, the revised guidance is "primarily due to higher component pricing and, to a lesser extent, additional data center costs to support future year capacity." Meta is deploying "more than one gigawatt" of custom silicon, developed with Broadcom, alongside AMD and Nvidia systems, all focused on AI model training and inference. This shift highlights a strategic pivot: while R&D spending also grew significantly, CapEx surpassed R&D in 2025 ($72.22 billion vs. $57.37 billion) and is projected to widen the gap further in 2026. This indicates that the foundational models for Meta’s AI vision, such as Llama, are largely developed, and the current bottleneck—and therefore the primary investment—is in deploying them at an immense scale.
Zuckerberg's vision is clear: to "deliver personal superintelligence to billions of people" through AI agents. This ambition necessitates colossal compute power. However, with direct AI monetization largely absent from Meta’s current revenue streams, this infrastructure arms race is a high-stakes bet. It demands that the underlying AI capabilities indirectly drive engagement and ad revenue, and eventually new proprietary services, to justify the immense capital deployment. The open-weight nature of Llama further complicates this, as Meta bears the costs for infrastructure supporting models that are also freely available to competitors.
Despite Meta’s strong operational performance, several significant risks suggest the market may be underpricing the downside to its aggressive AI strategy. The primary concern is the sheer scale of the CapEx burden versus uncertain monetization timing. The projected $125-$145 billion in 2026 CapEx, an 87% increase from 2025, is primarily driven by AI infrastructure costs, not near-term revenue streams. In Q1 2026, Meta's $19.84 billion in CapEx already outstripped its $12.39 billion in free cash flow, illustrating the immediate financial strain. Management has conceded that generative AI revenue will not be a meaningful driver in the next one to two years, leaving investors reliant on indirect benefits to justify the massive investment.
A persistent drag on profitability comes from Reality Labs' structural losses. In Q1 2026, RL reported an operating loss of $4.028 billion, offsetting 17.6% of Meta’s total operating income. For the full year 2025, RL losses of $19.19 billion reduced total operating income by approximately 23%. With RL investments projected to be similar in 2026, the mixed-reality/AI glasses segment continues to absorb billions with little revenue, creating a substantial hurdle for overall profitability, especially if AI monetization from other areas is delayed.
Furthermore, Meta faces escalating regulatory and geopolitical risks around AI and China. Mark Zuckerberg’s public opposition to U.S. bans on Chinese AI models, while consistent with his open-access philosophy, could spark political backlash in Washington. Meta is conspicuously the only major U.S. AI developer not to have signed a voluntary agreement with the Center for AI Standards and Innovation (CAISI) for pre-release model reviews, a framework designed to address national security concerns. This stance, coupled with Meta's refusal to sign the EU's General-Purpose AI (GPAI) Code of Practice and reported restrictions on Llama 4 variants in the EU, highlights growing regulatory friction. Potential compliance costs, estimated at $8-15 million annually for EU AI Act obligations alone, could add further financial pressure.
Finally, the erosion of Llama's open-weight dominance is a critical, underappreciated risk. While Meta has cultivated a vast developer ecosystem for Llama, Chinese models like Alibaba's Qwen have significantly outpaced Llama in downloads on platforms like Hugging Face, with Qwen nearly doubling Llama's cumulative downloads by March 2026. If Meta continues to bear the massive infrastructure costs for an open-weight ecosystem where its models are no longer the clear market leaders, it risks effectively subsidizing competitors without securing commensurately higher proprietary returns. The fact that 97.7% of Meta’s Q1 2026 revenue remains ad-driven further exposes the company to traditional advertising headwinds if the AI ROI story falters.
Wall Street analysts largely maintain a bullish stance on Meta Platforms, despite the significant capital expenditure ramp-up. Out of 65 analysts covering META, 50 recommend a Buy, and two a Strong Buy, leading to a consensus rating of Buy. The average 12-month price target stands at $804.17, with a median of $825.00. Optimists project a high target of $880.00, while even the most cautious set a low target of $700.00.
The following table summarizes the analyst price targets:
| Statistic | Value |
|---|---|
| Consensus Target | $804.17 |
| Median Target | $825.00 |
| High Target | $880.00 |
| Low Target | $700.00 |
| Current Price | $593.41 |
Against the current price of $593.41, these targets imply substantial upside, ranging from approximately 18.0% to the low target, to 35.5% to the consensus, and up to 39.0% to the median target. This reflects an underlying belief among analysts that Meta’s aggressive AI investments will translate into significant future earnings growth.
However, the consensus is not without its shifts. In early June 2026, Arete upgraded Meta from Neutral to Buy, signaling renewed confidence. Conversely, in late April 2026, JPMorgan downgraded the stock from Overweight to Neutral, adjusting its target after Meta's Q1 earnings and the raised CapEx guidance. This divergence highlights a key tension: while the market generally acknowledges Meta's long-term potential in AI, the scale and timing of the CapEx without immediate, direct monetization have prompted some to adopt a more cautious outlook. The wide spread between the low and high targets—a $180 difference, representing nearly a 26% variation from the low target—further illustrates the uncertainty regarding the return profile of Meta’s AI bet.
The most compelling counter-argument to a cautious stance on Meta is that the market is already pricing in a "capex risk discount" while simultaneously embedding high expectations for Meta's long-term AI monetization. Proponents of this view point to Meta's current trailing price-to-earnings multiple, which at approximately 21 times, sits about 25% below its 10-year average of roughly 28 times. This discount, they argue, reflects investor apprehension about the massive $125-$145 billion 2026 CapEx, suggesting the downside risk from high spending is already factored into the stock's valuation. Furthermore, analysts' long-term EPS forecasts, reaching $56.52 by 2030, imply a forward price-to-earnings ratio of only about 10.5 times against the current price, indicating that substantial AI-driven earnings growth is implicitly expected, but without an "AI bubble" multiple.
However, this objection, while powerful, only partially holds. While the market has indeed applied a discount relative to historical P/E, this doesn't fully account for the structural shift in Meta's AI strategy. The core problem is that Meta is bearing the full cost of an infrastructure build-out for open-weight models, whose market dominance is now actively being eroded by rivals like Alibaba's Qwen. Llama's loss of leadership in open-source downloads means Meta is effectively subsidizing an ecosystem from which it increasingly derives only indirect benefits, without the strong proprietary returns that could justify such a massive, sustained CapEx. The implicit expectation of substantial, direct new revenue streams from AI agents or enterprise licensing, which would de-risk the CapEx, has yet to materialize beyond ad-driven efficiencies.
Therefore, my thesis narrows: Meta’s substantial CapEx is not sufficiently de-risked by its open-weight strategy. The market's current targets, while reflecting long-term growth, may be overly optimistic about the quality of that growth given the diminishing proprietary returns from the open-source layer. The figure in the next report that would truly decide this debate, beyond the usual revenue and EPS, would be a clear, quantifiable metric for new, direct AI monetization streams—for instance, a specific revenue line item for Meta AI services, or a significant acceleration in non-advertising revenue growth that can be demonstrably linked to its AI agent ecosystem. Without this, the current CapEx remains an investment in an increasingly commoditized layer of the AI stack.
Meta Platforms, currently trading at $593.41, presents a complex investment proposition. Its aggressive $125 billion to $145 billion AI CapEx in 2026 is a statement of intent to lead the AI future, but this comes with significant risks. While Meta’s open-access philosophy has built a vast Llama ecosystem, its market share in open-weight models is visibly eroding to competitors like Qwen, and core monetization remains overwhelmingly ad-driven. This dynamic suggests Meta is investing heavily in infrastructure for an open ecosystem that it no longer fully dominates, creating a substantial CapEx overhang without clear proprietary returns.
Given these factors, my stance on Meta is Neutral. While the company's operational performance remains strong, the multi-billion-dollar CapEx commitment, coupled with the slow direct monetization of AI and the declining leadership of Llama in the open-source arena, suggests that the market’s current implied upside to analyst targets of 35% to 39% may be premature.
For investors considering an entry, a more prudent zone would be closer to the 52-week low of $520.26. This would provide a larger margin of safety against the CapEx burden and geopolitical uncertainties. My 12-month target for Meta is $710.00, slightly above the low analyst target of $700.00, reflecting some long-term AI potential but factoring in the execution and competitive challenges. This target implies an approximate 19.6% upside from the current price. The invalidation level for this thesis would be if Meta’s 2027 CapEx guidance exceeds $150 billion without a corresponding acceleration in non-advertising AI revenue growth, or if Reality Labs' annual operating losses continue to remain above $20 billion. In this high-stakes AI race, Meta's open-source bet requires more tangible, proprietary wins to justify its monumental spend.
Want deeper research on any stock? Try Kavout Pro for AI-powered analysis, smart signals, and more. Already a member? Add credits to run more research.
