Microsoft and Meta's quarterly reports addressed market focus issues: computing power expenditure and AI demand are both very strong!

According to the Wind Trading Platform, JP Morgan analyst Harlan Sur interpreted the latest financial reports from Microsoft and Meta, pointing out that the intensity of spending on AI infrastructure has clearly entered a new expansion cycle.

Both tech giants indicated in their latest financial reports that the tight supply of AI computing power will persist through 2026. Both companies' capital expenditures for the fourth quarter exceeded market expectations: Microsoft at $37.5 billion and Meta at $22.137 billion. Meanwhile, Meta raised its full-year capital expenditure forecast for 2026 to $125 billion, a year-on-year increase of 73%.

This indicates that, driven by the accelerated deployment of foundational models, AI agents, and commercial applications, the demand for computing power continues to exceed supply capacity, driving cloud computing and hyperscale enterprises to increase their investments. JP Morgan believes that under the current supply-demand dynamics, major tech companies still have room to increase capital expenditures, with investments focusing on data centers, servers, and network infrastructure, thereby boosting the performance of related semiconductor industry chains. This investment trend is expected to continue until 2027.

Supply Limits Become the Norm: Demand Gap Will Persist

Supply tightness has become the core bottleneck in the current construction of AI infrastructure. Both Microsoft and Meta pointed out in their latest earnings call that the demand for computing power continues to exceed supply capacity.

The structural reasons for the supply-demand imbalance lie in the accelerated deployment of foundational models, AI agents, and commercial applications, driving exponential growth in computational intensity. Meta revealed that its GPU cluster used for training generative advertising models has doubled in size and is further expanding to support the training of the new generation GEM model in 2026.

This ongoing supply shortage is expected to continue supporting high-intensity investments in data centers, servers, and network infrastructure between 2026 and 2027.

Custom Chip Development Becomes a Strategic Focus

In addition to continuing to procure AMD and NVIDIA GPUs, Microsoft and Meta are accelerating their custom ASIC chip strategy to improve energy efficiency and expand application scenarios.

On Meta's side, its self-developed chip project MTIA continues to iterate, currently supporting search engine inference and planning to expand to core ranking, recommendation training, and inference loads in the first quarter of 2026. JP Morgan pointed out that Meta's chip design partner Broadcom will benefit from this, with significant revenue growth expected from Meta by 2026.

Microsoft is focusing on optimizing the energy efficiency and economics of token processing, emphasizing the critical role of custom chips. According to JP Morgan's analysis, although Marvell Technology did not participate in Microsoft's previously released MAIA 200 chip, it is advancing the development of the next-generation MAIA 300 chip as planned, which is expected to enter mass production in the second half of 2026.

As the scale and complexity of models increase, the demand for computing power is growing exponentially. The capital expenditure guidance from cloud service giants continues to focus on investments in AI infrastructure. **The latest plans from Meta and Microsoft further confirm JP Morgan's judgment: investments in networking, custom chips (ASICs), and GPUs for computing and storage acceleration will maintain strong momentum in the medium to long term **