Source URL: https://www.tomtunguz.com/earnings-microsoft-2025-04-30/
Source: Tomasz Tunguz
Title: 100 Trillion Tokens
Feedly Summary:
“We processed over 100t tokens this quarter, up 5x year over year, including a record 50t tokens last month alone.”
If the market harbored any doubt for the insatiable demand for AI, this statement during Microsoft’s quarterly earnings yesterday, quashed it.
What could this mean for a run rate? Using some basic assumptions1, this implies :
Scenario
Model Mix Assumption
Est Monthly Rev, $m
Annualized, $b
Expensive
70% OpenAI, 20% Claude, 10% Other
525
6.3
Medium
65% OpenAI, 20% Claude, 15% Other
358
4.3
Low
60% OpenAI, 20% Claude, 20% Other
256
3.1
So AI is roughly between 13 to 28% of Azure revenue. Error bars here are quite large, though.
A major contributor to this increased demand is performance, especially with reasoning models.
Combined with some of the massive reductions in inference costs, especially with smaller models like the Phi-4 models that Microsoft released yesterday that are open source and small. The margins on AI inference should continue to surge.
“…our cost per token, which has more than halved.”
“You see this in our supply chain where we have reduced dock to lead times for new GPUs by nearly 20% across our blended fleet where we have increased AI performance by nearly 30% ISO power…”
Jevon’s Paradox in full force.
“The real outperformance in Azure this quarter was in our non AI business.”
This was a surprise, but it likely is the result of additional demands placed on adjacent systems. AI doesn’t exist in a vacuum. It needs databases, storage, orchestration, and observability to succeed.
“PostgreSQL usage accelerated for the third consecutive quarter… Cosmos DB revenue growth also accelerated again this quarter…”
A later quote within the analyst call reinforces this point, the database systems, Cosmos (a MongoDB-like document data store) & PostGres, Both of which are transactional databases.
100 trillion tokens up 4x y/y. Next year, could we see a quadrillion?
1 1:10 input-to-output token ratio; a model usage mix of 60-70% OpenAI, 20% Anthropic, remainder of other models ; and a 20% discount to public prices. See the work here
AI Summary and Description: Yes
Summary: The text presents insights into the exponential growth in AI token processing, primarily attributed to Microsoft’s performance in the Azure cloud ecosystem. The anticipated financial implications suggest significant revenue contributions from AI services, driven by cost reductions and enhanced model performance. This growth trajectory highlights critical considerations for infrastructure and cloud computing professionals, particularly regarding the integration of AI with existing systems.
Detailed Description:
The provided text outlines key insights from Microsoft’s quarterly earnings presentation concerning the dramatic increase in AI token processing demand—up 5x year-over-year. The implications for revenue and operational efficiency underscore significant trends relevant to professionals in AI, cloud computing, and infrastructure.
Key Points:
– **Token Processing Growth**: Microsoft processed over 100 trillion tokens this quarter, with the previous month alone accounting for 50 trillion tokens, indicating a robust demand for AI capabilities.
– **Revenue Projections**: Based on different scenarios regarding the model mix (i.e., OpenAI, Claude, and others), estimated monthly revenues from AI services on Azure range from $256 million to $525 million:
– Expensive mix (70% OpenAI): $525 million monthly / $6.3 billion annually
– Medium mix (65% OpenAI): $358 million monthly / $4.3 billion annually
– Low mix (60% OpenAI): $256 million monthly / $3.1 billion annually
– **AI’s Contribution to Azure**: AI services constitute approximately 13-28% of Azure’s revenue, based on the scenarios discussed. This highlights the strategic importance of AI within cloud services.
– **Cost Efficiency**: A noted reduction in cost per token, which has more than halved, suggests a trend towards optimizing AI inference costs. This is particularly relevant for efforts in enhancing margins and profitability from AI offerings.
– **Supporting Infrastructure**: The text emphasizes that AI applications require comprehensive infrastructure support, including databases and observability tools. The accelerated usage of PostgreSQL and the growth of Cosmos DB indicate that adjacent systems are experiencing increased demand due to AI-related workloads.
– **Future Outlook**: The mention of reaching potentially a quadrillion tokens in upcoming quarters suggests a compounding trajectory of growth, raising considerations for scalability and resource management in cloud services.
Overall, the report underscores not only the rapid growth and monetization potential of AI within cloud environments but also the essential infrastructure enhancements necessary to sustain this demand, offering critical reflections on the broader implications for AI security and infrastructure management.