Nexo Earn with Nexo
Google AI project delay rattles chip stocks, but data center spending keeps rolling

Google AI project delay rattles chip stocks, but data center spending keeps rolling

Nearly half of planned US AI data center mega-projects face delays, yet hyperscalers are doubling down with record capital commitments

The AI infrastructure boom has a timing problem. Chip stocks are feeling the pain of delayed Google AI projects, but the companies actually writing the checks for data centers haven’t flinched, continuing to pour record amounts of capital into buildouts that will take years to complete.

The delay problem is bigger than Google

Nearly half of planned 2026 US AI data center mega-projects are facing some form of delay, postponement, or outright cancellation. Of the anticipated 12-16 GW of data center capacity, only about 5 GW is under active construction as of mid-2026.

JPMorgan found that over 60% of data center capacity planned for 2027 hasn’t even started construction yet.

The culprits are the usual suspects in large-scale infrastructure: power grid constraints, permitting difficulties, and supply chain bottlenecks.

Advertisement

For chip manufacturers like NVIDIA, AMD, and Broadcom, these delays translate directly into uncertainty about near-term demand. If the data centers aren’t being built on schedule, the chips that would fill them aren’t being ordered on schedule either.

Alphabet’s $85B bet says otherwise

In early June 2026, Alphabet announced plans to raise approximately $85 billion in equity to fund AI-related data center construction. The raise includes a $10 billion commitment from Berkshire Hathaway.

Alphabet’s shares still dropped 3.9% following the announcement.

Alphabet isn’t alone in writing enormous checks. The collective AI infrastructure spending by major tech companies, including Amazon, Meta, and Microsoft, may exceed $650 billion in 2026.

A bifurcated market emerges

What’s developing is a two-speed market in AI infrastructure. On one track, you have the immediate reality of construction delays, permitting headaches, and power constraints creating genuine uncertainty about when capacity comes online. On the other track, you have the largest technology companies on earth committing record capital because they believe AI workloads will eventually consume every megawatt they can build.

What this means for investors

The gap between announced spending and actual construction creates a peculiar dynamic for anyone holding chip stocks or data center-adjacent equities. The $650 billion in planned spending acts as a floor for long-term demand expectations. But the fact that only about 5 GW out of 12-16 GW of planned capacity is actively being built means that floor is further away than many investors initially priced in.

Semiconductor investors need to watch two things closely. First, whether the power grid and permitting bottlenecks show signs of easing, which would accelerate deployment timelines and pull chip orders forward. Second, whether any of the hyperscalers actually reduce their capital expenditure guidance, which would be the real bearish signal. So far, none have.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

Google AI project delay rattles chip stocks, but data center spending keeps rolling

Google AI project delay rattles chip stocks, but data center spending keeps rolling

Nearly half of planned US AI data center mega-projects face delays, yet hyperscalers are doubling down with record capital commitments

The AI infrastructure boom has a timing problem. Chip stocks are feeling the pain of delayed Google AI projects, but the companies actually writing the checks for data centers haven’t flinched, continuing to pour record amounts of capital into buildouts that will take years to complete.

The delay problem is bigger than Google

Nearly half of planned 2026 US AI data center mega-projects are facing some form of delay, postponement, or outright cancellation. Of the anticipated 12-16 GW of data center capacity, only about 5 GW is under active construction as of mid-2026.

JPMorgan found that over 60% of data center capacity planned for 2027 hasn’t even started construction yet.

The culprits are the usual suspects in large-scale infrastructure: power grid constraints, permitting difficulties, and supply chain bottlenecks.

Advertisement

For chip manufacturers like NVIDIA, AMD, and Broadcom, these delays translate directly into uncertainty about near-term demand. If the data centers aren’t being built on schedule, the chips that would fill them aren’t being ordered on schedule either.

Alphabet’s $85B bet says otherwise

In early June 2026, Alphabet announced plans to raise approximately $85 billion in equity to fund AI-related data center construction. The raise includes a $10 billion commitment from Berkshire Hathaway.

Alphabet’s shares still dropped 3.9% following the announcement.

Alphabet isn’t alone in writing enormous checks. The collective AI infrastructure spending by major tech companies, including Amazon, Meta, and Microsoft, may exceed $650 billion in 2026.

A bifurcated market emerges

What’s developing is a two-speed market in AI infrastructure. On one track, you have the immediate reality of construction delays, permitting headaches, and power constraints creating genuine uncertainty about when capacity comes online. On the other track, you have the largest technology companies on earth committing record capital because they believe AI workloads will eventually consume every megawatt they can build.

What this means for investors

The gap between announced spending and actual construction creates a peculiar dynamic for anyone holding chip stocks or data center-adjacent equities. The $650 billion in planned spending acts as a floor for long-term demand expectations. But the fact that only about 5 GW out of 12-16 GW of planned capacity is actively being built means that floor is further away than many investors initially priced in.

Semiconductor investors need to watch two things closely. First, whether the power grid and permitting bottlenecks show signs of easing, which would accelerate deployment timelines and pull chip orders forward. Second, whether any of the hyperscalers actually reduce their capital expenditure guidance, which would be the real bearish signal. So far, none have.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.