Join us in returning to NYC on June 5th to collaborate with executive leaders in exploring comprehensive methods for auditing AI models regarding bias, performance, and ethical compliance across diverse organizations. Find out how you can attend here.


Espresso AI, a stealthy Silicon Valley artificial intelligence startup, has raised over $11 million in seed funding to bring the power of AI to perhaps the biggest challenge in enterprise computing today: reining in runaway cloud costs. The funding includes a seed round led by Daniel Gross and Nat Friedman, and a pre-seed round led by Matt Turck at FirstMark, with participation from industry leaders.

The company, which emerged from stealth today, has developed technology that uses advanced language models and machine learning to automatically optimize code and reduce cloud compute costs by up to 80%. Its initial product focuses on streamlining SQL queries for Snowflake, the popular cloud data warehousing platform.

“The opportunity is just enormous,” Espresso AI founder and CEO Ben Lerner said in an exclusive interview with VentureBeat. “Snowflake alone has $2 billion in annual revenue. If you look across data warehousing broadly, it’s certainly hundreds of millions of dollars in revenue for us, and billions in potential savings for customers.”

A brewing crisis in cloud costs

The move to the cloud has been a double-edged sword for enterprises. While cloud platforms provide unparalleled flexibility and scalability, they have also introduced new challenges around cost control and visibility. Many organizations now find themselves grappling with unexpectedly high bills and struggling to forecast and manage their spend.

VB Event

The AI Impact Tour: The AI Audit

Join us as we return to NYC on June 5th to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.

Request an invite

Data warehousing is a particular pain point. As companies consolidate data silos and launch new analytics and machine learning initiatives, data warehouses have become some of the biggest consumers of cloud resources. But optimizing these workloads for cost and performance is notoriously difficult.

“What we hear all the time from users is that Snowflake is their second biggest line item after AWS,” Lerner told VentureBeat. “And if you go to any Snowflake event, they really focus on two things: cost and performance.”

AI to the rescue

Espresso AI’s solution is to harness the power of large language models (LLMs), the foundational technology behind viral sensations like ChatGPT, to tackle the problem of code optimization. By training these models to deeply understand SQL queries and database architectures, Espresso AI has built a platform that can automatically refactor queries to make them more efficient.

Here’s how it works: Espresso AI plugs into a company’s existing Snowflake setup and continuously analyzes the queries being run against the data warehouse. Using a combination of natural language processing, program synthesis, and reinforcement learning, it identifies opportunities for optimization and rewrites queries on the fly to improve performance and minimize compute usage.

“The reason this is so powerful is that for a lot of existing applications, you need to have a human in the loop to check for accuracy,” Lerner explained. “When you’re optimizing code, you already know what you want it to do – just go faster. And so we can automatically verify the optimized code is correct.”

Setup is designed to be simple, with the ability to get up and running in under 10 minutes by just changing a single connection string. “It’s as easy as changing a URL,” said Lerner. “You point your BI and analytics tools to the Espresso endpoint instead of directly to Snowflake, and we handle the rest.”

Poised for growth

Espresso AI has already seen strong early traction, with multiple enterprise customers using its platform to optimize production Snowflake workloads. The company plans to use its funding to accelerate product development and go-to-market efforts.

While Snowflake is the initial focus, Espresso AI’s technology is extensible to any SQL data warehouse. Support for platforms like Databricks is on the near-term roadmap. Longer-term, the company envisions using its AI optimization engine to speed up compute across the stack, from data pre-processing to model training.

“It’s hard to put a dollar amount on what the world will look like if computers run 100 times faster,” said Mr. Lerner. “Everything’s going to go fast. We’ll be able to do more research, more machine learning. There are tons of constraints around compute today.”

Of course, delivering 100x speedups is easier said than done. While Espresso AI has shown impressive results in early customer deployments, achieving order-of-magnitude performance gains will require significant research breakthroughs. The company will also need to fend off competition from cloud providers themselves, who are heavily investing in cost management and optimization capabilities.

But if Espresso AI can deliver on even a fraction of its founding vision, the implications could be profound. With enterprises spending over $600 billion annually on cloud and on-prem compute, the market opportunity for AI-powered efficiency gains is massive.

In an era of belt-tightening and digital transformation, technologies that can drive meaningful cost savings without sacrificing performance will find an eager audience among CIOs. By bringing the power of AI to the unsexy but essential domain of code optimization, Espresso AI may be brewing up something truly disruptive.

If a cup of coffee is the price to pay for taming cloud costs, expect a lot more IT leaders to line up for a sip of Espresso AI.

link

By admin