Market not buying Meta’s AI story: The cheap valuation masks three unresolved challenges

TL;DR

• Bank of America raised its capital expenditure forecasts for Alphabet, Meta, and AWS, predicting their combined capacity could reach 57GW by 2027.
• Model breakdown shows Meta's implied value per GW of AI capacity is approximately $4 billion, far lower than the two major cloud providers.
• Meta appears the cheapest, but enterprise AI sales, power access, and customer willingness to pay have not yet been fully proven.

In its latest report, Bank of America raised its future capital expenditure and data center capacity forecasts for Alphabet, Meta, and Amazon AWS, and presented a valuation breakdown with stark contrasts: according to its model, the implied value of Meta per GW of AI capacity in its current stock price is only about $4 billion, far lower than Alphabet's approximately $110 billion/GW and Amazon's approximately $59 billion/GW.

The key takeaway from this report isn't who spends the most, but that despite both expanding AI data centers, the market assigns vastly different prices to the capacity of different companies. AWS and Google Cloud have mature cloud businesses capable of selling computing power to enterprise clients. Meta relies more on its advertising business, AI recommendation efficiency, and its still-early enterprise AI products, resulting in a lower value for its data center capacity reflected in its stock price.

For investors, AI capital expenditure ultimately must answer a practical question: can electricity, GPUs, and data center capacity translate into cloud revenue, enterprise AI service revenue, or higher advertising efficiency? Meta's discount reflects that this question hasn't been fully bought into by the market yet.

Big Three's 2027 Capacity Could Reach 57GW, CapEx Still Being Revised Upwards

According to Bank of America's forecasts, the capital expenditure expectations for Alphabet, Meta, and AWS for 2026 to 2027 have been broadly revised upwards. Specifically, Alphabet's 2026 CapEx expectation was raised from $187 billion to $195 billion, and its 2027 expectation from $257 billion to $290 billion. Meta's 2026 expectation went from $130 billion to $145 billion, and its 2027 expectation from $157 billion to $185 billion. AWS's 2026 expectation remained at $159 billion, while its 2027 expectation was raised from $196 billion to $230 billion.

These figures are closer to Bank of America's model predictions and are not all equivalent to company guidance. In official statements, Meta has previously raised its 2026 capital expenditure guidance to $125-$145 billion, while Alphabet's public guidance is around $180-$190 billion.

Corresponding to data center capacity, Bank of America estimates the combined capacity of the three companies at about 27GW by the end of 2025, rising to 39GW in 2026, and further to 57GW in 2027. In other words, approximately 30GW of new capacity will be added within two years.

Amazon is expected to add the most. From 2026 to 2027, AWS is projected to add about 15GW, Google about 9GW, and Meta about 6GW. AWS itself has a larger cloud infrastructure base; client demand, internal e-commerce, and AI services collectively absorb the capacity, hence its largest expansion scale.

Comparison of old and new capital expenditure forecasts for Alphabet, Meta, and AWS (2026-2028), showing the most significant upward revision for 2027.

Building the same 1GW capacity also comes with different costs. According to Bank of America's estimates, the cost per additional GW of capacity in 2026 is approximately $25 billion for Amazon, $37 billion for Google, and $45 billion for Meta. Amazon's cost is the lowest, mainly due to economies of scale and its self-developed chips. Meta's cost is the highest, more impacted by upfront construction investment and reliance on external GPUs.

This puts Meta in a more awkward position: it is not adding the most capacity, yet its per-GW construction cost is higher. If it fails to successfully generate enterprise revenue or clearly reflect this in advertising efficiency, the market will find it harder to assign a higher valuation to this asset class in advance.

Valuation Gap Widens: Meta Valued at Only $4 Billion per GW

Bank of America's valuation breakdown method first strips out the traditional business value of the three companies, then deduces the implied value the market assigns to their AI capacity.

After backward calculation based on multiples of core advertising and retail revenue for 2027, the implied valuation per GW of AI capacity for Meta is only about $4 billion. For Alphabet, it's about $110 billion/GW, and for Amazon, about $59 billion/GW.

This gap directly points to the different commercialization paths of the three companies. The market is more willing to treat Alphabet and Amazon's data center capacity as monetizable assets, but remains significantly cautious about Meta's AI capacity.

Comparison of implied valuation per GW capacity: Alphabet ~$110 billion, Amazon ~$59 billion, Meta ~$4 billion.

The capacity of AWS and Google Cloud is more easily linked to cloud revenue. According to Bank of America's model, AWS's cloud revenue per GW in 2026 is about $10.6 billion, and Google Cloud's is about $15.7 billion. The path to revenue from enterprise clients purchasing cloud computing power, AI training, and inference services is relatively clear.

Meta is different. It has a massive advertising business and AI recommendation systems, but its enterprise-level AI revenue is still in an early stage. Even if Meta accelerates the construction of AI data centers, the market will still ask: will this capacity primarily enhance its own advertising efficiency, or can it be sold externally like cloud providers?

If primarily used internally, the valuation method would be closer to advertising efficiency improvement rather than an independent cloud infrastructure asset. For Meta to achieve a higher per-GW valuation, it needs to demonstrate a clearer path for enterprise AI products, subscription revenue, or Business Agent sales.

Meta's Upside Hinges on Its Ability to Sell Capacity

In Bank of America's optimistic scenario, Meta's data center capacity could reach approximately 22.8 to 23GW by 2030. If 40% of this is used for enterprise AI sales, calculated at a revenue rate of $12 billion/GW, it corresponds to a potential enterprise revenue opportunity of about $110 billion.

This remains a model assumption, not a management target or a confirmed revenue opportunity. It explains the source of the "Meta is undervalued" narrative: if Meta can productize some of its AI capacity in the future, selling AI services, subscription products, or Business Agent capabilities to enterprises, then the current implied value of ~$4 billion/GW seems very low.

Forecast installed capacity growth for Amazon, Google Cloud, and Meta (2026-2030). Meta's capacity is projected to be approximately 22.8GW by 2030.

The problem is that this assumption has not yet materialized. AWS and Google Cloud already have clients, contracts, and cloud revenue metrics. Meta needs to prove it is not just "building computing power for itself" but can also generate sustainable enterprise AI revenue.

Potential catalysts listed in the report include improvement in cloud gross margins, increased visibility for Meta's enterprise AI and subscription products, and more detailed disclosure of AI revenue breakdowns. Some longer-term products and partnerships remain more hypothetical and cannot be directly counted as realized business contributions.

Estimated Cloud/AI revenue per GW (2026-2030): AWS ~$10.0-10.6 billion, Google Cloud ~$15.2-17.0 billion, Meta conservative estimate ~$12 billion.

For Meta, the real game-changer is not announcing another larger data center plan, but showing investors what revenue this capacity can generate. Specifically, the proportion of enterprise AI sales, product forms, and revenue disclosure are currently unclear.

The Cheapest Asset Also Needs to Prove Itself the Most

Meta appears to have the cheapest valuation for its AI capacity among the three companies, but cheapness alone is not an answer.

The first constraint is electricity. A U.S. Department of Energy page previously cited an EPRI estimate that data center power consumption could account for up to about 9% of total U.S. electricity consumption by 2030, up from about 4% in 2023. More recent research ranges from EPRI and Lawrence Berkeley National Laboratory are higher, suggesting electricity pressure may continue to rise. Power access, transmission, local permitting, and energy prices will all affect whether planned GW capacity comes online on time.

The second constraint is chip availability and construction delivery. GPU supply, networking equipment, power infrastructure, and construction timelines all impact production schedules. Raising capital expenditure does not mean capacity is immediately online, nor does it mean revenue is immediately recognized.

The third constraint is customer willingness to pay. Enterprise AI demand is still growing, but the scale of revenue clients are willing to sustain for inference, training, and agent services requires further verification through financial reports. For Meta, if enterprise AI revenue is not clearly disclosed, the market will find it difficult to value its data center capacity by the same standards as cloud providers.

Therefore, the conclusion of Bank of America's report is not that "Meta has already realized its AI value," but rather a more direct valuation contrast: amidst all three internet giants continuing to expand AI capital expenditure, the market assigns the lowest price to Meta's data center capacity. It also has the most to prove, both in building the capacity and convincing investors that this capacity can translate into visible revenue.