Goldman Sachs Forecasts $765 Billion AI Spending by 2026 as Cash Flows Dwindle; Amazon Capex Raised to $220
Goldman Sachs forecasts that six major tech companies will spend $765 billion on AI in 2026, with total rising to nearly $1.2 trillion in 2027. Amazon raised its full-year capex forecast to $220 billion, while its free cash flow turned negative $7.6 billion. Meta's cash generation fell 91% year-over-year, and Alphabet posted its first quarter of negative free cash flow. Memory chip shortages are driving cost increases, with Tesla CEO Elon Musk calling memory pricing 'crazy' and Amazon CEO Andy Jassy citing 'price increases' for memory chips as a factor in capex guidance.
The AI boom is entering its fourth year, and global big tech companies continue to make grand commitments, but at the cost of accelerating cash burn. According to forecasts, the six major tech giants will spend a total of $765 billion on AI in 2026, rising to nearly $1.2 trillion in 2027. In July 2026, Amazon raised its full-year capital expenditure forecast to $220 billion, the highest among the four major cloud service providers, while its free cash flow over the past 12 months was negative $7.6 billion. Earlier, Meta disclosed that its cash generation capacity fell 91% compared with the same period last year, while Alphabet posted its first quarter of negative free cash flow, putting financial pressure on one of the world's most profitable companies. One of the core reasons for the surge in costs is a shortage of memory chips. Tesla CEO Elon Musk described memory pricing as 'crazy' during the company's earnings call, while Amazon CEO Andy Jassy said that 'price increases' for memory chips had pushed up the capex guidance. Market reactions to the earnings reports were mixed. Tesla and Alphabet saw their stock prices fall after reporting negative cash flow and accelerated spending. Apple's shares declined as memory shortages weighed on its outlook. Amazon, however, rose on strong growth in its cloud business. Still, apart from memory supplier Micron, none of the large-cap tech stocks achieved breakout gains in 2026, despite healthy revenue growth. The overall market performance was tepid, reflecting growing investor skepticism about whether debt-driven large-scale AI buildouts will ultimately generate returns. Meanwhile, Chinese AI labs have recently been releasing new models in rapid succession, narrowing the performance gap with OpenAI and Anthropic at lower prices, catering to the trend of US corporate clients increasingly focusing on cost control. These so-called open-weight models can be downloaded, modified, and deployed on any infrastructure.