Capital flows into "iron" and clouds

Major tech companies' investments in AI are approaching $650 billion in 2026, up from $410 billion a year earlier. It is no longer a matter of developing individual software solutions, but building large-scale infrastructure. The main recipients of capital are chip makers and cloud operators: NVIDIA, Broadcom, Microsoft, Alphabet, Amazon, and Meta.

Investors view them as participants in a capital-intensive cycle. Return on invested capital comes to the fore. Quotes respond to slowing orders or changing margins. Even with moderate growth in the S&P 500, the dynamics within the tech sector remain uneven.

Computing and energy as a finite resource

The expansion of AI infrastructure increases costs for data centers, network equipment, and specialized chips. In parallel, electricity consumption is growing. Computing power and energy are becoming basic planning factors. The demand for electricity generation and distribution is captured in long-term investment programs.

The macroeconomic background remains mixed, with job growth in the US expected to fall below January’s levels, from around 60,000 to 130,000 earlier. The unemployment rate is projected to be close to 4.3%. The contribution of AI to GDP is estimated at about 0.5 percentage points in 2025, with a possible increase to 1 percentage point in 2026. Productivity growth is accompanied by a redistribution of employment, as automation outstrips the formation of new positions.

Shifting focus: from model scale to standards

After the step of actively scaling up models, attention shifts to their applied efficiency.

Google DeepMind, Anthropic, OpenAI and Mistral AI research units are expanding their work in this direction. New laboratories and projects are being launched with separate funding.

One area is the standardization of agent systems. Support from the Linux Foundation and the creation of the Agentic AI Foundation are associated with the development of a universal protocol for data interaction models and APIs. Unification reduces integration costs for corporate users and increases competition between platforms.

The current phase of AI development involves a reallocation of capital to companies that control computing power, energy resources, and technology standards. The key market issue remains payback of $650 billion and growth in demand for cloud and AI services.