Will Alphabet’s TPUs Offset Its Declining Search Business?

Alphabet (Nasdaq: GOOG) recently announced its third quarter results that failed to meet market expectations. The performance sent the stock tumbling 13%, the worst drop since it went public in 2004.

Alphabet’s Financials

Alphabet’s third quarter revenue grew 25% to $127.6 billion, below analyst estimates of $127.9 billion. EPS of $2.12 was also short of analyst estimates of $3.01.

Among key segments, revenues from Google Services were up 11% to $96.3 billion. Search revenues grew 12% to $63 billion, missing the market estimates of $65 billion. While search query growth rate was the highest in the last five years, analysts are concerned that the lower than expected revenue is a symbol of AI cannibalizing search. Google attributed the growth in query rate to the success of its AI offerings such as AI Mode and AI Overviews.

Google Cloud revenues grew 103% to $30.7 billion compared with the Street’s estimates of $29.1 billion. Google Cloud now has a backlog of $460 billion. Revenues from Other bets were down to $372 million. Overall advertising revenues were up 11% to $82 billion, compared with estimates of $83.5 billion. YouTube advertising revenues grew 9% to $11.2 billion.

Capex spending continues to go up for all companies. Alphabet is no anomaly. It increased its earlier estimated capital expenditure outlay for the year from $180-$190 billion to $195-$205 billion. It incurred $51.4 billion in capex in the third quarter of the year.

Alphabet did not provide an estimate for the fourth quarter or the full year. Analysts are looking for revenues of $125.33 billion for the fourth quarter and an EPS of $3.26. For the full year, analysts expect revenues of $435.4 billion and an EPS of $20.60.

Alphabet’s AI Infrastructure Focus

Google attributed its Cloud segment growth to the huge demand for its AI infrastructure and AI solutions. It is seeing strong demand for its Tensor Processing Units (TPUs) across customers. The AI offerings had very strong operating metrics. Its first-party model APIs are processing nearly 32 billion tokens per minute. The Gemini app crossed one billion users, becoming the fastest-growing product in its history.

Google is increasing its footprint within the AI infrastructure marketing through its custom TPUs, Axion CPUs, and the latest NVIDIA GPUs. NVIDIA GPUs are a core part of Google’s AI accelerator portfolio. Through its partnership with NVIDIA, Google will be among the first to offer NVIDIA Vera Rubin NVL72, along with the Blackwell- and Hopper-based instances that are already available.

Google also released its eighth generation TPUs that are specialized for training and serving and can take on demanding agentic workloads. TPU 8t comes with three times the processing power of Ironwood and two times the performance to power AI research. Google is seeing positive market reactions to its products. Oracle recently announced that it will become the first company to use Google’s AI processors for its own data centers.

Within Gemini, its 3.1 Pro continues to expand its reasoning and multimodal understanding capabilities. 3.1 Pro now also includes MCP support and native visualizations. The cost-efficient Gemini Flash model, 3.1 Flash Live, has improved audio capabilities, higher precision and reasoning and has voice interactions that are more natural and intuitive. Speech-to-text capability is now available in 70 languages.

Google is also integrating AI capabilities across its products and platforms. Earlier this month, it integrated Nano Banana 2 to make personalized image creation possible in the Gemini app. It also embedded AI capabilities within Maps so that users can get more personalized suggestions and intuitive directions.

Despite its moves, the market is not pleased with Google. Its core Search business performance has been particularly disappointing with this being the second consecutive quarter that it missed analyst expectations for the segment. Google’s search queries may be increasing, but its AI offerings are not driving monetization. Additionally, there is rising competition from players like OpenAI, which is working on an ads business for ChatGPT, and Meta, which is using its agent Muse to lure ad dollars. Google will need to figure out how to accelerate monetization from its AI Mode and AI Overview features within Search, if it wants to placate the market.

Google’s stock is trading at $282.32 with a market capitalization of $3.42 trillion. It hit a 52-week high of $404.47 in June and has recovered from the 52-week low of $271.41 that it was trading at in April.