Should Google monetize its AI accelerator chip business? DA Davidson weighs in

Alphabet has introduced Ironwood, its seventh-generation Tensor Processing Unit (TPU), which analysts at DA Davidson view as a rapidly closing alternative to Nvidia's dominant AI chips, noting double the energy efficiency over its predecessor. Despite being currently limited to Google's internal use or cloud services, DA Davidson identifies significant external demand, particularly from cutting-edge labs, and estimates a standalone TPU and DeepMind business could be valued at $900 billion. However, given the unlikelihood of a spin-off that would unlock this value, the firm maintains a 'neutral' rating on Alphabet, raising its price target to $190 pending consideration of a breakup.

Analysis

Alphabet's introduction of its seventh-generation Tensor Processing Unit (TPU), 'Ironwood', marks a significant advancement in its proprietary AI hardware capabilities. According to analysts at DA Davidson, the new chip doubles the performance-per-watt over its predecessor and is substantially closing the competitive gap with market leader Nvidia. This technological progress is corroborated by a 96% growth in developer activity for TPUs on Google Cloud between February and August. The core of the investment thesis, however, lies in the valuation disconnect; DA Davidson estimates a hypothetical standalone business combining TPUs and DeepMind could be worth approximately $900 billion, a notable increase from a prior $717 billion estimate. This suggests substantial value is currently obscured within Alphabet's conglomerate structure. Despite identifying significant latent demand from external labs like Anthropic and xAI, the TPUs remain captive to Google's internal use and its cloud platform. Consequently, while analysts raised their price target to $190, they maintain a 'Neutral' rating, citing the low probability of a near-term spin-off required to unlock this embedded value.

Key Decisions for Investors

• Investors should recognize the significant embedded value within Alphabet's AI hardware division, but acknowledge that its realization is contingent on a corporate spin-off or change in sales strategy, which analysts currently view as unlikely.
• Monitor any management commentary or strategic shifts regarding the external sale of TPU systems, as such a move would be a primary catalyst for a stock re-rating and unlocking the estimated $900 billion valuation.
• While Alphabet's TPU progress solidifies its competitive standing against Nvidia, its immediate financial impact is confined to enhancing Google Cloud's offerings rather than generating direct hardware revenue, a key distinction for valuation models.