Gold prices tick higher on fresh U.S. tariff threats, Fed rate cut hopes
Investing.com -- Concerns about Alphabet (NASDAQ:GOOGL) losing market share to ChatGPT and the rising costs of AI in search have been at the center of investor debates recently.
In a new report released Monday, Barclays (LON:BARC) analysts weighed in on these issues, focusing on Google’s recent disclosure of AI inference tokens.
According to Barclays, Alphabet is currently processing 480 trillion tokens per month across its products and APIs, up sharply from 9.7 trillion a year earlier.
The report states that this nearly 50-fold increase is "likely a function of AI Overviews in Search primarily, other AI search features like Lens and Circle to Search, and third-party developers.”
By comparison, Microsoft (NASDAQ:MSFT) recently disclosed that it inferenced 100 trillion tokens in the first quarter of 2025, with 50 trillion of those processed in March alone.
Barclays estimates that Google “is processing 5-6x more tokens than Microsoft (Azure), as Search is ~6x the size of ChatGPT, which is ~2-4x the size of Gemini."
The analysis also addresses cost implications, estimating that Alphabet spent about $750 million on inference tokens in the first quarter of 2025. The bank notes that “the rate of deleverage from infusing AI into Search appears manageable (which may come as a surprise to some investors).”
The AI Overview costs represent roughly 1% of Search revenue, compared to core costs running at around 18% of revenue, excluding traffic acquisition costs.
Looking ahead, token consumption is expected to keep rising. The analysts highlight that agents—AI systems capable of handling more complex queries—are beginning to drive further token usage.
Moreover, several forthcoming Google products, including Project Astra, Project Mariner, and Veo, have yet to be fully released and could contribute to even higher AI workloads.
“Our estimate for GOOGL’s compute capex now accounts for over half of total capex spend, and it continues to tick upwards,” the report notes.
Despite the rapid growth in AI usage, Barclays observes that there is “no impact” on Google’s company-level operating income (OI) margin yet, with token costs accounting for only about 1.6% of costs of goods sold (COGS) and operating expenses in the first quarter.
Annualized, the $750 million quarterly token cost would represent around 1.4% of 2025’s Search revenue. Even assuming most tokens are tied to Search and AI Overviews, AI costs likely account for about 1% of Search revenue.
Barclays concludes that while there is some margin pressure, it remains relatively small. However, with token volumes rising sharply, cost management will be important as AI usage expands.