Krishna Rao: Effective compute procurement is vital for tech success, dynamic resource allocation boosts model efficiency, and shifting to exponential growth is essential for innovation | Invest Like the Best

Strategic compute resource management is driving exponential growth and unlocking new revenue streams in AI development.

Listen on Invest Like the Best with Patrick O'Shaughnessy

Share

Add us on Google by Editorial Team May. 13, 2026

Key takeaways

• Effective compute procurement is vital for tech companies, balancing between excess and scarcity to avoid business failure.
• Utilizing multiple chip platforms enhances flexibility and efficiency in resource management.
• Collaboration with chip manufacturers can maximize the value of compute investments.
• Exponential growth in tech requires a shift from linear to non-linear thinking.
• Strategic compute allocation is essential for balancing model development and customer service.
• Dynamic compute resource allocation improves model development efficiency and cost-effectiveness.
• New AI model iterations consistently show significant improvements in capability and efficiency.
• Advancing model intelligence opens up new business opportunities and use cases.
• Significant leaps in model intelligence can dramatically increase company revenue.
• Scaling laws in AI continue to drive accelerated progress in model development.
• The multidimensional nature of intelligence unlocks new enterprise applications.
• Model intelligence advancements directly correlate with business growth and revenue increases.
• Effective resource management strategies are crucial for the rapid evolution of AI technologies.
• Partnerships with technology developers enhance operational efficiency and innovation.
• Understanding the balance between resource allocation and business objectives is key to success in tech.

Guest intro

Krishna Rao serves as Chief Financial Officer of Anthropic. He previously served as CFO of Fanatics Commerce and secured a $1 billion lifeline for Airbnb during the 2020 pandemic. At Anthropic since May 2024, he oversees financial strategy amid the company’s rapid scaling to a $30 billion annualized revenue run rate.

The critical role of compute procurement

• The compute that we procure is the lifeblood of our business

  — Krishna Rao

• Effective compute procurement is crucial for balancing business needs and avoiding failure.
• Mismanagement of compute resources can lead to business failure, either through excess or scarcity.

• If you buy too much compute you go out of business if you buy too little compute you can’t serve your customers

  — Krishna Rao

• Strategic planning is essential to optimize compute procurement and allocation.
• Compute resources serve as the foundation for building technological capabilities.
• The balance in compute procurement directly impacts customer service capabilities.
• Understanding the significance of compute in AI and tech companies is crucial for strategic planning.
• The consequences of mismanagement in compute procurement can be severe for tech businesses.

Flexibility in chip platform usage

• We use three different chip platforms so we are customers of amazon’s trainium chip google’s cpus and nvidia’s gpus

  — Krishna Rao

• Utilizing multiple chip platforms allows for flexible and efficient resource management.
• The ability to use various chip platforms enhances operational efficiency.

• That flexibility took us a long time to be able to do that

  — Krishna Rao

• Strategic use of multiple platforms optimizes compute resources.
• Flexibility in chip usage is a key component of effective resource management.
• Collaboration with chip manufacturers enhances the value of compute investments.

• A dollar of compute inside our organization goes further than I think it does anywhere else

  — Krishna Rao

• Partnerships with technology developers are crucial for maximizing compute value.
• Investing in flexibility is essential for optimizing technological capabilities.

Shifting from linear to exponential growth

• I’ve been at the company for two years that’s something that’s paradigm I’ve had to break for myself

  — Krishna Rao

• Exponential growth requires a shift from linear to non-linear thinking.
• The range of outcomes in exponential growth is significantly wider than in linear growth.
• Understanding the difference between linear and exponential growth is crucial for business strategy.
• Navigating growth in rapidly evolving industries requires a mindset shift.

• When you’re on this exponential again the range of outcomes starts to be really really wide

  — Krishna Rao

• Exponential growth presents unique challenges and opportunities for businesses.
• Strategic planning must account for the non-linear nature of exponential growth.
• Businesses must adapt to the dynamic nature of exponential growth to remain competitive.

Strategic compute allocation for development and service

• There’s a level of compute for model development that we will not go below

  — Krishna Rao

• Strategic compute allocation balances model development and customer service.
• Compute resources are critical for both internal development and external service.
• Prioritizing compute for model development is essential for long-term success.
• Balancing compute allocation is a strategic decision with significant business implications.

• Even if it means it’s harder to serve customers or we have to do kind of unnatural things

  — Krishna Rao

• Effective resource management is key to achieving business objectives.
• Understanding the impact of compute resources on development and service is crucial for strategy.
• Businesses must prioritize compute allocation to support both development and customer needs.

Dynamic allocation for model development efficiency

• We can allocate that compute so dynamically we can make changes

  — Krishna Rao

• Dynamic compute allocation enhances model development efficiency.
• Adjustments in compute allocation can be made on a short time horizon for improved performance.
• Dynamic resource management leads to cost-effectiveness in model development.

• The efficiency thing is so interesting to me

  — Krishna Rao

• Efficient resource allocation is crucial for advancements in AI technology.
• Dynamic allocation strategies improve both capability and efficiency in model development.
• Businesses benefit from the flexibility of dynamic compute resource management.
• Understanding the significance of dynamic allocation is key to optimizing model development.

Advancements in model intelligence

• We think of it differently intelligence for us is multidimensional

  — Krishna Rao

• Advancing model intelligence unlocks new business opportunities and use cases.
• The multidimensional nature of intelligence enhances enterprise applications.

• By releasing new models the tam is unlocked in a unique way

  — Krishna Rao

• Model intelligence advancements directly correlate with business growth and revenue increases.
• Significant leaps in model intelligence can dramatically increase company revenue.

• We started the year with about $9,000,000,000 of run rate revenue

  — Krishna Rao

• Understanding the impact of model intelligence on financial performance is crucial for strategy.
• Businesses must prioritize advancements in model intelligence to remain competitive.

The impact of scaling laws on AI development

• We do see progress accelerating

  — Krishna Rao

• Scaling laws in AI continue to drive accelerated progress in model development.
• Understanding the implications of scaling laws is crucial for AI development.

• For us the scaling laws are alive and well

  — Krishna Rao

• Scaling laws influence model performance and technological advancements.
• Businesses must adapt to the rapid progress driven by scaling laws in AI.
• The ongoing advancements in AI present both challenges and opportunities for industry stakeholders.
• Strategic planning must account for the impact of scaling laws on AI development.
• Businesses benefit from understanding the role of scaling laws in technological evolution.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.