Key facts
- Google is developing a new server chip, 'Frozen v2,' to enhance Gemini AI model efficiency.
- The chip is projected to be six to 10 times more efficient than current Google AI chips.
- The 'Frozen v2' chip is potentially slated for a 2028 launch.
- Alphabet faces investor concerns over AI spending and a delay in the Gemini 3.5 Pro model launch.
- The company has raised its 2026 capital expenditure guidance to $180-$190 billion.
- Alphabet's cloud sales are expected to grow significantly, driving overall revenue.
Alphabet is reportedly developing a new server chip, codenamed 'Frozen v2,' aimed at improving the efficiency of its Gemini AI models. The chip, potentially launching by 2028, is expected to be six to 10 times more efficient than Google's current AI hardware in terms of tokens generated per unit of power. This initiative aligns with a broader industry trend of AI companies developing custom silicon to address computing capacity shortages and reduce reliance on Nvidia.
However, Alphabet faces investor scrutiny as the launch of its flagship Gemini 3.5 Pro model has been delayed, raising concerns about the payoff from significant data-center spending. The company has increased its 2026 capital expenditure guidance to between $180 billion and $190 billion. While Alphabet shares have lagged other 'Magnificent Seven' stocks recently, analysts anticipate a slight easing in revenue growth for the second quarter, primarily driven by its cloud business, which has benefited from deals involving Google's custom AI chips.
The company has also seen key employees depart for rivals. Analysts suggest Google's strength lies in its consumer distribution network alongside its AI models, infrastructure, and custom silicon. The drive for more efficient AI hardware is becoming critical as market euphoria around AI cools and companies focus on demonstrating returns on substantial investments.