👩🍳 How we use AI at Tech in Asia, thoughtfully and responsibly.
🧔♂️ A friendly human may check it before it goes live. More news here
🧔♂️ A friendly human may check it before it goes live. More news here
Citigroup sees record M&A in 2026 driven by AI
Citigroup forecasts a record year for M&A activity in 2026, driven by AI despite recent volatility in tech stocks and broader markets.
Senior bankers noted that recent sell-offs across tech, commodities, and cryptocurrencies may create winners and losers, but overall valuations remain healthy.
The bank highlighted a pipeline of AI-related deals, including Singapore’s S$13.8 billion (US$10.8 billion) acquisition of ST Telemedia Global Data Centres by KKR and Singtel, as the largest M&A deal in four years.
Experts expect 2026 to be the strongest year ever, supported by economic growth, low inflation, and abundant private equity capital.
🔗 Source: South China Morning Post
🧠 Food for thought
Implications, context, and why it matters.
The M&A boom is hiding a capital gap for AI infrastructure
- Private equity cash can look plentiful, yet the spending needed for data centers and other compute gear for AI remains far larger.
- Bain estimates that meeting AI compute demand by 2030 needs about $500 billion in capital investment each year, which then needs $2 trillion in annual revenue to support it 1.
- Even under optimistic assumptions, Bain still calculates an $800 billion annual revenue shortfall, and says the gap is far beyond any expected government subsidies 1.
- Dealmaking picked up in 2025, yet M&A funding hit a 10-year low, only 7% of cash spending across almost 700 S&P World Index companies, as budgets shifted to tech stacks, robots and AI, factories, and energy farms 2.
Power and land are turning into deciding factors in the AI economy
- The S$13.8 billion ST Telemedia Global Data Centres deal focuses on locking up sites for physical infrastructure with reliable access to power.
- The purchase gives KKR and Singtel control of a platform with more than 100 data centers and 2.3 gigawatts of total design capacity, as AI compute demand grows more than twice the rate of Moore’s law (a long-running observation that chip performance tends to improve rapidly over time) 3, 1.
- Constraints are already tight, and US demand alone could reach 100 gigawatts by 2030, adding pressure to a power grid that has seen relatively flat load growth for the past 20 years 1.
- Companies that secure grid connections, land, and permits can gain leverage, and access to AI may depend on electricity supply as much as software 4.
