Meta Stock Rises on a Hidden AI Infrastructure Catalyst
Meta’s new Muse Spark 1.1 AI model has attracted plenty of attention. However, Wall Street appears to be focused on a very different reason for the company’s latest share-price rally.
The stronger catalyst is buried inside Meta’s infrastructure plans.
According to details first reported by Reuters, an internal company memo suggests Meta may be building its artificial intelligence capacity far more efficiently than analysts previously expected.
The report prompted BofA Securities analyst Justin Post to maintain a Buy rating on Meta Platforms and keep his $835 price target.
Investors Reassess Meta’s AI Spending
For months, investors have worried about Meta’s enormous capital expenditure budget.
Building advanced AI systems requires vast amounts of computing power, electricity, data centres and specialised hardware. As a result, many analysts feared that Meta’s spending could place heavy pressure on cash flow.
However, the internal memo suggests the company is not simply expanding aggressively. It may also be reducing the cost of every unit of computing capacity.
That change could significantly improve the economics of Meta’s AI strategy.
Meta Plans a Massive Increase in Computing Capacity
Meta reportedly aims to add approximately 14 gigawatts of total computing capacity across 2026 and 2027.
The company had already deployed around 1 gigawatt in 2026, according to the memo. It also expected to add another 5.5 gigawatts during the second half of the year.
These figures are far above earlier Wall Street estimates.
BofA had previously expected Meta to add about 2.6 gigawatts of capacity. If the internal projections are accurate, Meta’s expansion is happening much faster than analysts anticipated.
Meta’s Cost per Gigawatt May Be Far Lower
The biggest surprise is the estimated cost of that expansion.
BofA had previously assumed that Meta would spend around $45 billion for every gigawatt of new computing capacity.
However, the capacity figures in the memo, combined with Meta’s expected $145 billion capital expenditure budget, suggest the real cost could be closer to $22 billion per gigawatt.
That would mean Meta is building its AI infrastructure at roughly half the cost Wall Street expected.
Lower Infrastructure Costs Could Transform Meta’s AI Returns
The main concern surrounding Meta’s AI investment has been the lack of a clear and immediate return.
Many investors feared that the company would spend enormous amounts of money without generating enough revenue to justify the investment.
BofA’s analysis challenges that view.
If Meta can build computing capacity for less than $30 billion per gigawatt, the financial returns could be highly attractive compared with other major technology companies.
Post noted that Amazon and Google may generate around $10 billion to $16 billion in annual cloud revenue per gigawatt. He also compared Meta’s costs with recent capacity agreements linked to SpaceX, which could range between $40 billion and $50 billion per gigawatt each year.
These comparisons suggest Meta may be creating unusually strong economics around its AI infrastructure.
Meta’s Iris AI Chip Adds Another Long-Term Catalyst
Reuters also reported that Meta plans to begin manufacturing a custom chip known as Iris in the fall.
The chip is expected to support Meta’s existing purchases of graphics processing units and reduce its dependence on external suppliers over time.
Meta is reportedly working with Broadcom and TSMC to begin producing Iris in September.
However, BofA does not believe the custom chip is responsible for the company’s current cost savings.
Since Iris will not enter production until later in the year, Meta’s 2026 infrastructure efficiencies appear to be coming from other operational improvements.
That makes the reported cost reductions even more significant.
Meta Plans a Broader Custom-Chip Roadmap
The Iris chip may still become an important long-term advantage.
Meta reportedly plans to introduce new custom chips roughly every six months through 2027.
The company is also securing long-term supply agreements with key manufacturing partners, including Broadcom and TSMC.
A successful custom-chip strategy could lower hardware costs, improve efficiency and give Meta greater control over its AI infrastructure.
It could also strengthen profit margins in a similar way to Google’s TPU chips and Amazon’s Trainium processors.
The Real Reason Meta Shares Are Surging
Most headlines are focused on Meta’s latest AI software.
However, the company’s rising share price appears to be driven more by the economics of its infrastructure strategy.
Wall Street is beginning to see evidence that Meta is not spending blindly in an attempt to compete with OpenAI and other AI leaders.
Instead, the company may be building a highly efficient and vertically integrated AI platform.
By potentially cutting computing-capacity costs in half and developing its own chip pipeline, Meta is addressing one of investors’ biggest concerns.
The company is starting to show that its massive AI capital expenditure budget could eventually produce strong and sustainable returns.
