Alphabet shares rose nearly 3% after reports emerged that the company is developing a new AI chip, Frozen v2, designed to run its Gemini AI models with up to ten times greater efficiency.
The Frozen v2 chip will embed parts of the Gemini model directly into silicon, reducing data transfer and lowering power consumption per AI token processed. Google engineers estimate it could handle six to ten times more tokens per unit of power compared to their latest tensor processing units (TPUs). By contrast, Google's current seventh-generation TPU, Ironwood, only doubled the efficiency of its predecessor.
Scheduled for launch in 2028, the new chip will support existing TPUs dating back to 2016 but depends on Google maintaining Gemini’s core design to achieve these efficiency gains.
Alphabet’s urgency to boost AI compute capacity has increased amid reported shortages that have forced Google Cloud to reject business. Google recently agreed to pay nearly $1 billion monthly to SpaceX to address this gap, while competitors like Amazon, Microsoft, and Meta are also investing in custom AI chips.
The announcement helped Alphabet recover some losses after a steep selloff in semiconductor stocks last week, with the VanEck Semiconductor ETF falling 8.9%, marking its worst performance since April 2025. The Philadelphia Semiconductor Index dropped nearly 10% amid fears of an AI market shakeout sparked by new AI models from startups and profit-taking from major investors.
Despite the volatility, Wall Street analysts view the selloff as a buying opportunity. Morgan Stanley’s Joseph Moore and Mizuho’s Vijay Rakesh highlighted sustained AI capital expenditures and power infrastructure growth driving long-term semiconductor demand. JPMorgan’s Mislav Matejka also expects buyers to return, noting new chip supplies will not arrive before 2028.
This material is for informational purposes only and does not constitute financial advice.
