Archive · 11 Jan 2026
Ajay Bagga Warns of Risks as Hyperscalers Invest Trillions in AI
Ajay Bagga, a market expert, has warned about the risks and changes happening as big tech companies spend a lot of money on artificial intelligence (AI).
These companies, called hyperscalers, are planning to spend $450 billion in 2025 and even more in 2026.
This big spending is changing software companies into ones that need a lot of money for hardware.
Bagga is worried about things like not knowing how much money these companies will make, how quickly technology becomes old, and how to make sure AI is fair and understandable.
He also talked about how these companies are making deals to get lots of electricity for their AI and data centers, even teaming up with nuclear power companies.
All of this shows how important energy is for AI and how tech companies are doing things they didn't do before.
Hyperscalers are expected to spend $450 billion in 2025 and $600 billion in 2026 on AI infrastructure.
This spending is transforming software firms into capital-intensive hardware operators.
Concerns include uncertainty over revenue visibility, faster technology obsolescence, and evolving depreciation standards.
Power generation, transmission, and storage are critical enablers of AI and data center rollout.
Meta has struck long-term power purchase deals with nuclear power companies to support its AI ambitions.
- Who
- Ajay Bagga, a market expert, and global technology companies (hyperscalers)
- What
- Hyperscalers are investing heavily in AI infrastructure, transforming software firms into capital-intensive hardware operators
- When
- The spending is expected to reach $450 billion in 2025 and $600 billion in 2026
- Where
- Globally, with a focus on AI-driven infrastructure and energy supply deals
- Why
- To expand AI-driven infrastructure and secure long-term energy supply for data centers
- Hyperscalers' Expected Spending in 2025
- $450 billion
- Hyperscalers' Expected Spending in 2026
- $600 billion
- Key Concerns
- Uncertainty over revenue visibility, faster technology obsolescence, commoditised pricing, and evolving depreciation standards
- Critical Enablers of AI and Data Centre Rollout
- Power generation, transmission, and storage
- Meta's Long-term Power Purchase Deal
- 20-year deal with Vistra
- Other Power Supply Agreements
- Commitments with Oklo and TerraPower for small modular nuclear reactors
- Expected Timeline for New Nuclear Reactors
- Early 2030s