Alphabet nears Nvidia in market value on AI demand

Alphabet briefly overtook Nvidia in after-hours market value this week after strong quarterly results.

A report said Anthropic planned to spend US$200 billion on Google Cloud over five years.

Alphabet ended the week at US$4.8 trillion, behind Nvidia at US$5.2 trillion, and its shares have risen about 160% in the past year as investors warmed to its AI and cloud businesses.

The company said its cloud backlog nearly doubled to US$462 billion, but the increase may be heavily tied to Anthropic, creating customer concentration risk, said Gil Luria, analyst at D.A. Davidson.

Mizuho estimated about US$61 billion of the backlog through 2027 could come from Google’s tensor processing units.

Some analysts warned its planned capital spending of up to US$190 billion could pressure returns.

🔗 Source: CNBC

🧠 Food for thought

Implications, context, and why it matters.

Anthropic and OpenAI deals now make up about half of big cloud providers’ backlogs

Reliance on a small group of customers is not limited to Google. It runs across the largest cloud providers 1.
Anthropic is an AI startup. OpenAI makes ChatGPT. Their contracts add up to about US$1.05 trillion. That equals roughly half of the US$2.1 trillion backlog held by the four largest U.S. cloud providers 1.
This wave of infrastructure spending depends on two AI startups with deeply negative free cash flow. That creates an unusual financial tie for this tech cycle 1.

Rising costs are pushing Google to sell more AI infrastructure outside cloud rentals

Spending on data centers and chips has reached the biggest infrastructure buildout in tech history. The top four U.S. tech firms expect a combined US$725 billion in capital expenditures for 2026 1.
Those costs are pushing companies to look beyond cloud rentals for returns 1.
Alphabet plans to start third-party Tensor Processing Unit (TPU) delivery for select customers’ data centers. That would let some customers use Google’s custom AI chips in their own facilities 2.
Most of that hardware revenue is expected in fiscal year 2027. The timing signals a change in how Google makes money from its AI infrastructure 2.