Why We Had been First Cash Into Blitzy


Right this moment, Blitzy introduced a $200M spherical at a $1.4B valuation, led by Northzone, PSG, and Battery Ventures.

Blitzy is an autonomous software program growth platform constructed for the Fortune 500. The place AI coding instruments like Cursor assist particular person builders write code quicker, Blitzy delivers total enterprise software program tasks, ingesting current codebases (100M+ traces), orchestrating 1000’s of AI brokers to autonomously write, take a look at, compile, and validate code, and transport 80%+ of what would in any other case be multi-month engineering tasks. (People polish the remaining work).

NFX wrote the primary test into the corporate in 2024, when the founders have been nonetheless MBAs at Harvard Enterprise Faculty. This is the reason we wrote that first test:

1. A $1.6 trillion companies market, prepared for AI disruption

Blitzy goes after the worldwide IT companies market, a $1.6 trillion+ category. Customized enterprise software program growth is among the largest slices inside it, and CIOs have traditionally outsourced most of that to huge contractor armies.

In 2024, Blitzy acknowledged that enterprise software program was the proper market to disrupt with autonomy. And so did we. Across the identical time, we wrote our AI Workforce thesis, predicting that trillions of {dollars} in companies income would turn into addressable by AI-native corporations for the primary time.

Blitzy was doing precisely that in one of many largest TAM expressions we’ve seen thus far.

Since then, we’ve seen the workforce remedy key execution challenges. They’ve managed to promote into quite a few Fortune 500 corporations. Convert $250k pilots into 7-figure annual contracts, and rework their prospects’ software program growth lifecycles. They’ve confirmed to be a wonderful mixture of imaginative and prescient and execution.

2. A founder pairing constructed for the enterprise

Promoting autonomous software program growth into the Fortune 500 requires each deep technical credibility (as a result of CIOs will dissect your structure) and operator polish (due to enterprise procurement cycles). Brian and Sid have been the founder pairing we had been on the lookout for for the reason that class emerged.

Brian is a West Level graduate, former US Military Ranger, Harvard MBA, and serial entrepreneur; the credible operator who can stroll right into a Fortune 500 CIO’s workplace and shut a seven-figure pilot.

Sid is a former NVIDIA Grasp Inventor with 27+ patents in generative AI and an early adopter of AI fashions for software program growth, the uncommon CTO who noticed the subsequent structure years earlier than the market.

3. A contrarian architectural guess — now vindicated

In 2024, the consensus on AI coding for the enterprise was easy: higher basis fashions would remedy the issue. Brian and Sid disagreed. Their thesis was that enterprise software program growth wasn’t simply an intelligence downside; it was an issue of complexity, belief, and programs.

Actual enterprise codebases span a whole lot of hundreds of thousands to hundreds of thousands of traces of code, with a long time of architectural selections and undocumented dependencies that no mannequin’s context window can seize. To truly make autonomous software program work on this context, you want a persistent information graph of the codebase plus a hyperscaled orchestration layer that routes throughout the very best frontier fashions.

In different phrases, it’s essential to understand how the codebase is designed and linked, and break that logic into manageable duties for the AI to disentangle and reassemble in your goal future state. That’s the way you really design software program that works in a real-world context.

They have been proper. Their system is so good at understanding the context round an organization’s codebase that it’s now #1 on SWE-Bench Pro at 66.5%. It’s higher at fixing real-life software program growth and execution issues within the enterprise than the newest releases from OpenAI, Anthropic, and Google.

That’s the facility of contrarian considering, a robust thesis, and glorious execution.

TLDR;

We have been first in with Blitzy as a result of;

  • The corporate was attacking one of many largest companies markets on the earth
  • As a result of Brian and Sid had the uncommon mixture of operator polish and technical depth to drive affect throughout the Fortune 500
  • Their architectural guess was contrarian (and concurrently made sense from a primary ideas perspective)

Two years later, that is apparent considering. However again then, they have been really seeing what others couldn’t. We’re proud to have backed them from the very starting.

Congratulations to Brian, Sid, and your complete Blitzy workforce.