Tech Giants Plan $720 Billion AI Spend Despite Election Uncertainty

Major technology companies are accelerating their spending on artificial intelligence infrastructure.

Amazon, Microsoft, Alphabet, and Meta plan to spend at least $720 billion on capital expenditures this year. Most of this funding will go toward building AI data centers. This figure does not include at least $1 trillion in spending commitments for future years.

Executives at these firms told investors they intend to continue this level of investment. Alphabet plans to spend between $195 billion and $205 billion in 2026. CFO Anat Ashkenazi said the company expects to increase spending in 2027. Amazon CEO Andy Jassy stated the company will spend "meaningfully more" next year than its planned $220 billion this year. He called the expansion a "once in a lifetime opportunity" to grow Amazon into a $1 trillion company. Meta expects to spend between $135 billion and $145 billion this year. Financial institutions predict this number will rise in 2027. Microsoft posted $45.2 billion in capital expenditures for the last quarter, a figure lower than the $50 billion analysts expected.

Investors reacted sharply to these financial plans. Alphabet stock dropped 13% in a single day last week, marking its largest one-day decline in history. The selloff in tech stocks began to taper off as investors turned their attention to Tuesday’s midterm elections. Some candidates have proposed moratoriums on data centers. Wall Street appears to view this potential regulatory change as a positive development for stock prices.

Despite high spending, the companies have yet to generate substantial profits from AI. Google Search ad revenue grew 12% in the third quarter, down from a 17% growth rate in the same period last year. Philipp Schindler, Alphabet’s senior vice president, acknowledged that the growth in AI Mode queries negatively impacted the core Search business. He noted that monetization for the new AI features remains lower than for traditional search ads. Jassy defended the spending timeline, stating that spreading capital expenditures over 25 years would be more convenient but is not the reality of the market.

Microsoft became the only hyperscaler to take on debt for its AI buildout last week. The company announced a $25 billion bond sale split into 10 parts. CFO Amy Hood said on a Wednesday earnings call that the company had not yet tapped the debt markets. That changed by the following morning when the bond sale news broke. Microsoft attributed its lower quarterly spending to "third party capacity delays" rather than cost-cutting measures.

Anthropic, an AI startup, updated its S-1 prospectus with third-quarter financial data. The company reported $17 billion in revenue, a more than 1,000% increase from last year. However, its compute costs reached $17.5 billion, leaving it with a net loss. Anthropic has committed nearly $900 billion to future compute and infrastructure costs. Part of this amount will go to Google, which sells TPU chips to Anthropic. Google executives said the company is capacity-constrained, yet it continues to sell its own hardware to other labs.

Mark Zuckerberg, Meta’s CEO, said the company expects compute efficiency to increase tenfold next year. He noted that this improvement will require significant investment. Zuckerberg also stated that Meta’s agentic AI program will only become profitable if the company reduces its compute costs.