Varun

Co-Founder of GigaML and CEO & Co-Founder of Windsurf (formerly Exafunction)

Claims

As speaker

Mentioned in

Key positions and views

Pivoting decisively based on market signals, even at the cost of existing revenue, is essential for survival and success in the AI industry.

The primary value of AI in software engineering is to massively multiply the productivity of existing developers, not to enable non-technical users to code.

The market for generic LLM fine-tuning services is inherently weak because its use cases are limited to incremental cost and speed improvements rather than creating new capabilities.

The AI model provider market is not analogous to the cloud market (AWS, GCP) as it lacks significant customer lock-in and has low switching costs, making it less defensible.

Enterprise AI adoption is currently bottlenecked by complex configuration and workflow integration challenges, creating a major opportunity for products that can automate this 'forward deployment' process.

Podcast consensus on Varun

Points of consensus

▶Varun consistently emphasizes a product strategy centered on solving high-value enterprise problems, evidenced by GigaML's focus on customer support for clients like DoorDash and Windsurf's focus on developer productivity for clients like JPMorgan Chase.May 2026

▶He believes in the transformative power of AI to dramatically increase operational efficiency, claiming AI coding agents reduce his engineering needs by 6-7 times at GigaML and that Windsurf's mission is to make engineers 10x more productive.May 2026

▶Across his ventures, Varun demonstrates a pattern of rapid pivoting based on market feedback, moving from edtech to customer support AI at GigaML, and from GPU virtualization to developer tools at Windsurf.May 2026

▶He has successfully leveraged specific growth tactics for different stages, using open-source model releases to gain initial traction and raise a seed round for GigaML, and later developing high-performance proprietary models like SWE1 for Windsurf's enterprise product.May 2026

Points of debate

▶Varun's approach to product development appears to vary significantly between ventures. GigaML's process requires securing a paying customer commitment before building, whereas Windsurf executed a pivot over a single weekend, abandoning millions in revenue from a previous business.May 2026

▶His company's core focus shows a strategic split. GigaML is dedicated to building AI agents for customer support, while Windsurf's primary mission is to augment the productivity of existing software engineers.May 2026

▶There is a contrast in his market-entry strategy. GigaML gained its initial traction and funding through popular open-source models, while Windsurf's current competitive advantage is tied to its proprietary, high-performance SWE1 model.May 2026

▶His perspective on market viability seems to have evolved. He founded Exafunction on the premise that many model architectures would proliferate, a bet that failed, while his analysis of the LLM fine-tuning market for GigaML is much more critical and narrow, focusing only on cost and speed.May 2026

Key themes

▶Pragmatic Pivoting and Market Adaptation

Varun's career is defined by decisive pivots away from failing ideas toward market-validated opportunities. He abandoned GigaML's initial edtech concept and Exafunction's GPU virtualization business, which was based on a flawed market prediction and faltering customer base, in favor of customer support AI and developer tools, respectively.

This demonstrates a highly adaptable, non-ideological approach to company building, suggesting an ability to quickly cut losses and reallocate resources toward product-market fit, a key trait for navigating the volatile AI landscape.

▶AI for Extreme Productivity GainsMay 2026

A core belief in Varun's strategy is using AI to create order-of-magnitude improvements in efficiency. This is evident in his claim that GigaML would need 6-7 times more engineers without AI coding agents and Windsurf's stated goal of making engineers 10x more productive, positioning AI as a force multiplier rather than a replacement.

Investors should note his focus on quantifiable, high-impact ROI for both internal operations and customer value propositions, which is a strong selling point for enterprise adoption.

▶Enterprise-Focused AI Product DevelopmentMay 2026

Both GigaML and Windsurf are oriented towards solving complex enterprise problems. GigaML targets high-volume customer support deflection for clients like DoorDash and major telecom providers, while Windsurf sells its developer tools to large organizations like JPMorgan Chase.

Varun's focus on the enterprise segment, characterized by high-value contracts and complex workflows, indicates a strategy geared towards building defensible, deeply integrated products rather than competing in the more commoditized consumer or prosumer AI markets.

▶Technical Moats through Performance and Integration

Varun emphasizes building competitive advantages through superior technical performance. For Windsurf's autocomplete, he notes that every 10ms of latency reduces acceptance by 1%, driving the development of the proprietary SWE1 model and even a custom IDE to overcome VS Code's limitations.

This focus on deep technical optimization, from model performance to UI latency, suggests a product strategy that competes on user experience and efficiency, creating a sticky product that is difficult for competitors to replicate without similar engineering investment.

Source episodes

Timeline

Pre-Founding

Varun earned approximately $50,000 from Kaggle competitions and turned down a $550,000 job offer from a quantitative trading firm to start a company.

Exafunction Founding

Founded Exafunction to build GPU virtualization technology, based on the incorrect premise that hundreds of different AI model architectures would become popular. Raised a Series A from Greenoaks under this model.

GigaML Founding & YC

Applied to Y Combinator with an initial idea for an edtech product using LLMs, which was discouraged by the partners.

Mid-2022

Triggered a major pivot at Exafunction when its primary customers (autonomous vehicle companies) began failing to raise new funding rounds. The company pivoted to the Codium product (now Windsurf) over a single weekend, abandoning its previous business.

GigaML Pivot & Seed Round

GigaML gained significant traction by open-sourcing models that topped Hugging Face benchmarks, leading to a $4 million seed round. The company pivoted its focus to customer support AI after observing it was a top use case among its fine-tuning customers.

Present

Leads GigaML, which serves major clients like DoorDash, and Windsurf, which serves clients like JPMorgan Chase and operates its own proprietary model (SWE1) processing hundreds of billions of tokens daily.

Suggested prompts

How does Varun reconcile the strategy of securing customer commitment before building (GigaML) with the rapid, revenue-abandoning pivot executed at Windsurf? &nearr;Given his experience with both open-source traction (GigaML) and proprietary models (Windsurf's SWE1), what is his framework for deciding when to open-source versus when to build closed IP? &nearr;How did the failure of Exafunction's core market assumption—that diverse AI architectures would proliferate—inform his current views on the defensibility of AI model providers? &nearr;Varun claims AI agents will soon handle the entire software development lifecycle. What specific technical or adoption hurdles does he believe must be overcome to make this a reality within his 12-month prediction? &nearr;

Key concepts

Company Pivots 2 ep AI Product Strategy 2 ep Developer Productivity Tools 1 ep Customer Support Automation 1 ep Fundraising and Venture Capital 2 ep Proprietary AI Models 1 ep Market Analysis 2 ep Enterprise Sales 2 ep Open Source Strategy 1 ep

Notable quotes

“We on the other hand decided we wanted to work on Codium over a weekend, me and my co-founder, and we told the company on Monday, and everyone started working on it that Monday, despite the fact that we were making a couple million in revenue off the previous business.”

Varun · Windsurf CEO & Co-Founder, Varun Mohan: AI's Biggest Acquisition to Date!

“I think there should be at least like 6 to 7 times more compared to now.”

Varun · Why Two IIT Engineers Turned Down $550K Jobs To Build A Startup

“Even for a tab product, 10 milliseconds, which is the product that gives you not only the autocomplete, but the automatic refactors as the user's typing, every 10 milliseconds affects a percentage point in acceptance rate right now for the product.”

Varun · Windsurf CEO & Co-Founder, Varun Mohan: AI's Biggest Acquisition to Date!

“I mean, Anthropic and OpenAI doesn't even pay salespeople commissions.”

Varun · Why Two IIT Engineers Turned Down $550K Jobs To Build A Startup

Report last updated: Jun 10, 2026

Get started free

Back to Entities Intelligence Report