To The Moon

OpenAI Turns to TPUs: What Does This Mean for Google?

On Saturday, The Information reported that OpenAI, one of the world's largest AI chip customers, has recently begun leasing Google's TPU chips to power products like ChatGPT. This marks the first time the company has used non-Nvidia chips on a large scale.

OpenAI primarily leases NVIDIA server chips through Microsoft and Oracle to develop and train models and provide computational power for ChatGPT. It's understood that the company spent over $4 billion on such servers last year, with training and inference costs each accounting for half. The company's expenditure on AI chip servers is expected to approach $14 billion by 2025.

The direct driving force behind the shift to Google's TPUs was the viral success of ChatGPT's image generation tool earlier this year, which put enormous pressure on OpenAI's inference servers at Microsoft. To address the growing computational demands and cost pressures, OpenAI first sought support from Google Cloud.

What Does This Mean for Google?

Google being one of the earliest and most technologically advanced in this field. OpenAI's shift towards Google's TPU chips may also indicate new changes in the AI infrastructure landscape.

Morgan Stanley believes that OpenAI is by far the most significant TPU customer (other customers include Apple, Safe Superintelligence, and Cohere), and this agreement is a crucial endorsement of Google's AI infrastructure capabilities. Google's TPU technology has been evolving for a decade, with the initial TPU released in 2015.

Analysts suggest this collaboration brings two major positive impacts for Google: First, it could become a driver for accelerated growth in Google Cloud revenue, a factor not yet reflected in GOOGL's stock price. Second, it demonstrates Google's confidence in its position in the search domain.

This indicates a significant opportunity for Google in terms of market share shift or TAM expansion. If OpenAI leads more customers to migrate, Google Cloud's Compute TAM could be rapidly revised upward.

Notably, despite not having access to the most advanced versions, OpenAI's choice to use TPUs further affirms Google's leading position in the broader ASIC ecosystem. As developers become more familiar with TPUs, increased adoption by companies outside of Google could serve as an additional growth driver for Google's cloud business.

Despite Lowest PE Among M7, Morgan Stanley's TP Remains Cautious

In recent years, Google's valuation has been the lowest among the M7, trading at around 19 times earnings, due to market concerns that generative AI could disrupt its search engine business (both in terms of market share and profit margins). 

However, this significant endorsement of Google's infrastructure capabilities by OpenAI is likely to drive growth in Google's cloud business and may partially alleviate market worries about AI's adverse effects on the company.

While Morgan Stanley sees marginal improvement in fundamentals, they remain cautious on the price target, maintaining their previous target of $185. This represents only a 4% upside from last Friday's closing price.

In contrast, Goldman Sachs took a more optimistic stance in their report a month ago, reaffirming a "Buy" rating with a 12-month price target of $220. Goldman believes that Alphabet's competitive advantages are underestimated in the long term. 

They note that Google continues to accelerate innovation in its core search business, integrating AI into multiple search distribution points and formats (such as AI Overviews, AI Mode, Gemini standalone chatbot, and Gemini Live). Goldman Sachs expects Google to maintain healthier revenue growth than investors currently anticipate.

@TigerStars  @CaptainTiger  @TigerWire  @Daily_Discussion  @Tiger_chat  @Tiger_comments  @MillionaireTiger