Summary

• GTLB is taking a defensive stance toward the rise of AI IDEs and coding agents, underestimating their sophistication and growth potential.
• The company’s platform expansion strategy has been broad but poorly executed, lacking the focus and discipline seen at peers like PANW and DDOG.
• Leadership assumes that more code generation will translate into more GitLab usage, but stagnant developer hiring challenges this logic.
• GTLB remains reliant on legacy enterprise customers while faster-moving digital-native firms increasingly adopt AI-native development tools.
• Without a bold pivot into AI IDEs and agentic workflows, GTLB risks being sidelined as the developer stack is reshaped by AI-first competitors.

GTLB's Passive Response

If you look closely at the evolution of AI applications — spanning AI IDEs, agents, code generation, and now AI coding agents — you’ll see that Windsurf, Cursor, and Replit have moved far beyond simply serving as copilots or acting as resellers for foundation models. To truly make foundation models (FM) excel in real-world software development, these startups have built their own copilots and crafted extensive logic to augment the base models. They’ve also accumulated vast amounts of developer feedback data, invaluable for training next-generation AI coding agents. This data covers everything from developers’ preferred debugging and testing strategies to their approaches for fixing and validating code — ensuring that AI codegen can reliably output functioning code, even for massive codebases with millions of lines.

In short, the code generation capabilities of foundation models have improved drastically over the past few quarters. Meanwhile, the upper layers — AI-powered IDEs — have also advanced rapidly, reaching a point where true mass adoption is underway. These platforms are proving that they’re not merely passing along API calls to the FM, but are building a critical, value-added stack that amplifies the FM’s performance and makes the vision of AI coding agents increasingly real.

So, where does GTLB stand amid this AI revolution?

It may not be fair to blame the ex-CEO and founder, who was fighting cancer during this period of dramatic transition. But for the new CEO, William Staples — previously at the helm of New Relic — the response has been underwhelming. We’re concerned that leadership’s approach to AI feels shallow, self-protective, and overly path-dependent — a mindset more fitting for a legacy company coasting on past successes than for a software leader with ambitions to shape the future.

It appears the board is primarily focused on ensuring the new CEO continues GTLB’s platform expansion strategy across all aspects of CI/CD and DevOps workflows. They’re also intent on guiding GTLB’s shift from a seat-based business model to a consumption-based model — an approach that may offer more security in an AI-powered future where the growth in developer headcount could slow or even turn negative, limiting seat expansion opportunities. But there’s a difference between being secure and being a winner. What we’re seeing from management is a desire to safeguard the company’s position, not to aggressively pursue disruption or leadership in the AI era.

More concerning, we’re also seeing either clues of denial or attempts at suppressing investor fears disingenuously. Staples has tried to convince the investor community that the skyrocketing usage of AI IDEs will simply lead to more software creation, which in turn means GTLB will have more software to manage and hence more revenue to capture. This framing could be true, but it is aggressively presumptive — it assumes AI IDEs will stop short of aggressively expanding into CI/CD and adjacent DevOps domains. Moreover, it’s not clear whether AI is actually generating more software in aggregate, or instead mainly improving the quality of software via robust code generation, debugging, and testing. The reality is that AI’s benefits today seem split between quantity and quality, and arguably skew more toward the latter. Better, more streamlined debugging and testing should indeed accelerate the DevOps cycle and deliver more software over time, but for now, GTLB’s narrative feels more like wishful extrapolation than a forward-leaning strategy for leadership in the AI era.

In summary: GTLB is not well-positioned to benefit from the AI tailwinds sweeping through the industry. Rather than seizing the moment, it’s taking a passive stance — downplaying the sophistication and value-add of these new AI-powered platforms and reassuring itself that it is “safe and secure” in the face of disruption, instead of boldly trying to win the future.

Morgan Stanley Technology Conference 2025 March

"Ultimately, I see software creation layer as more of a commodity." The GTLB CEO believes VS code is really not making money at all, and VS Code plus GitHub are more of a GTM booster than a direct revenue contributor for MSFT.

1Q25 ER call

The current GitLab (GTLB) CEO believes that AI codegen copilots will expand the pool of developers and increase the amount of code written, which in turn should benefit GTLB as a downstream player in the software lifecycle. His argument, as reflected in his public remarks, is that more code means more need for platforms like GitLab to manage, test, secure, and deploy that code. He downplays the value and sophistication of AI IDEs, suggesting they’re more of a commodity layer, and reiterates that GitLab also offers AI codegen through its own Duo product. In his view, GTLB’s “defensive mode” is to focus on the lifecycle management — security, compliance, collaboration — rather than compete head-on with the frontier codegen and agent tools.