23.07.2026 - Capex: The AI Race is getting Expensive
Last night, Alphabet became the first of the major hyperscalers to report second-quarter earnings, providing investors with an important first look at the economics behind the AI investment cycle.
Hyperscalers: the world's largest cloud computing companies, operating massive global data center networks capable of scaling computing power almost without limits. The leading hyperscalers include Alphabet (Google Cloud), Microsoft (Azure), Amazon (AWS) and Meta, all of which are investing hundreds of billions of dollars into AI infrastructure.
Alphabet delivered another strong quarter. Revenue exceeded expectations, supported by an impressive 82% year-over-year increase in Google Cloud revenue, while operating income remained solid.
However, one number overshadowed everything else.
The company raised its expected 2026 capital expenditure (capex) to USD 195–205 billion, up from the previous guidance of USD 180–190 billion. Second-quarter capex alone doubled from a year ago to USD 44.9 billion, reflecting the enormous investments required for AI infrastructure.
Although the operating business remains highly profitable, Alphabet's capital expenditures now exceed its quarterly operating income. As a result, free cash flow turned negative, highlighting how aggressively cash is being reinvested into the AI race.
Markets:
Equities: Global markets are down led by technology stocks, with Nasdaq Futures down -1.5%; Google is down -7%
Bonds: yields continue to climb, US 10y yield around 4.71%, US 30y at 5.19%, Japan 10y yield 2.78%
Commodities: Oil prices jumps more than 5%, with WTI around USD 91/barrel and Brent around USD 100/barrel
Precious metals: fall, gold at USD 4’050/oz, silver trades above USD 57/oz
Currencies: USD is rising
Cryptos: falling slightly - Bitcoin at USD 65k
Volatility: jumps above 19, however still low
My View: Last night's results confirm a concern I have highlighted for quite some time. The AI race is becoming a capital expenditure race.
Alphabet is spending enormous amounts of cash simply to remain competitive. The hope is that these investments will eventually generate attractive returns. I remain increasingly skeptical.
Today, the company is spending more cash than it generates from its operations. That leads to negative free cash flow. While this is not unusual during periods of heavy investment, it is still a warning sign that investors should not simply ignore.
The common counterargument is straightforward: if investment spending slows in the future, free cash flow will recover quickly. That is true in theory.
The problem is that the money has already been spent. Even more importantly, can these companies realistically reduce CapEx? Every major technology company fears falling behind in AI. The competitive pressure has created an environment where spending is almost mandatory. It increasingly resembles a winner-takes-all race in which nobody feels able to step off the accelerator.
Another factor deserves close attention.
Chinese AI companies continue to demonstrate that highly competitive models can be developed with significantly lower investment budgets than their US counterparts. If similar performance can ultimately be achieved with a fraction of the capital, investors should begin asking difficult questions about whether the current spending levels are economically justified.
Regular readers know my position on the broader AI cycle. I believe artificial intelligence will become a valuable productivity and supporting tool across many industries, knowing that the perfect tool without any mistakes remains wishful thinking.
I remain far less convinced that today's unprecedented spending will ultimately generate returns that justify the hundreds of billions of dollars currently being invested.
The technology itself is not my concern, the economics behind the AI race are.
Become a member to access more valuable market updates like this
