_____ ___ ____ ___ __ __ __ __ _____ ____ ___ _ _____ _____ ____ _____ _ _____ _ ____ _____ | ___/ _ \| _ \ |_ _| \/ | \/ | ____| _ \_ _| / \|_ _| ____| | _ \| ____| | | ____| / \ / ___|| ____| | |_ | | | | |_) | | || |\/| | |\/| | _| | | | | | / _ \ | | | _| | |_) | _| | | | _| / _ \ \___ \| _| | _|| |_| | _ < | || | | | | | | |___| |_| | | / ___ \| | | |___ | _ <| |___| |___| |___ / ___ \ ___) | |___ |_| \___/|_| \_\ |___|_| |_|_| |_|_____|____/___/_/ \_\_| |_____| |_| \_\_____|_____|_____/_/ \_\____/|_____|
Citigroup’s Global Technology Banking Chief Philip Drury Joins AI company poolside as Chief Investment Officer
Phil Drury, head of Global Technology, Media, Communications Banking at Citigroup, has left to join poolside, a frontier AI lab building its own foundation models to achieve artificial general intelligence (AGI). Phil joins poolside as its first Chief Investment Officer. Phil brings to poolside 29 years’ experience having advised on some of the world’s most notable IPOs and mergers worth hundreds of billions of dollars. The move reflects the growing recognition that AI infrastructure —requiring billions in capital for GPUs and data centers —has become a new frontier for major dealmaking.
Drury’s appointment bolsters poolside’s leadership team as artificial intelligence transitions from experimental technology to societally-critical infrastructure. Training frontier models requires tens to hundreds of thousands of GPUs, gigawatts of power, and billion-dollar data centers, with annual compute costs reaching hundreds of millions. His mandate centers on structuring the complex financial architecture required for AI infrastructure while expanding poolside’s relationships with customers and investors.
“Phil brings deep dealmaking expertise and relationships across tech and finance that will accelerate our mission,” said CEO Jason Warner (ex-CTO, GitHub). “His decision to join poolside validates our approach to building human-level AI through software engineering, and his expertise will help us get there faster.” The appointment underscores poolside’s unique position in the artificial intelligence landscape. The company has built the largest environment for coding synthetic data, generating its own training material at scale—reaching trillions of synthetic tokens—through a system that learns from its own code execution. This self-improving approach reduces reliance on the web-scraped content that the industry has rapidly depleted.
“The AI opportunity is increasingly about building the right infrastructure,” said Drury. “Building human-level AI through software engineering requires both technical excellence and strategic capital allocation. I’m excited to help poolside accelerate this mission by connecting our technology with the resources and partners who can help us get to AGI.”
Over Drury’s 29-year tenure at Citigroup he has held various banking leadership roles across products and geographies, including Head of Americas Equity Capital Markets; Head of EMEA Banking, Capital Markets & Advisory; and, since 2021, Head of Global Technology, Media & Communications. Drury’s track record includes advising on several of the world’s largest IPOs including Saudi Aramco ($25bn), Alibaba ($21bn), Visa ($17bn), AIA ($17bn), Facebook ($16bn) and GM ($15bn). More recently Drury and his team have advised on many of the most high profile combinations across TMT aggregating nearly $1 trillion, and raising over $270 billion in equity capital and $2.5 trillion in debt capital on behalf of clients.
Media Contact: press@poolside.com
About poolside
poolside is a foundation AI lab on the path to unlocking human-level artificial intelligence by building sophisticated models for software engineering. Founded in 2023 and headquartered in San Francisco, we deploy models directly into the development environments of government, defense, and other highly-regulated organizations. Using our Reinforcement Learning from Code Execution Feedback approach, our foundation models continuously improve by learning from successes and failures—just like human developers do, but at machine scale.