Rivalry is heating up and putting software services in the firing line

Its impetus stems from public disclosures of executive political affiliations and campaign funding, alongside broader unease over what critics see as a shift towards more closed research practices.

OpenAI is formidable by any objective metric. As of early 2026, it reported an estimated 810 million monthly active users. At this scale, even marginal defections matter. Some 700,000 subscribers have reportedly signed up to ‘quit’ its ChatGPT and other tools.

The deeper significance of this movement, however, lies not in its numbers, but in the structural conditions that make exits feasible. A capability convergence among large language models (LLMs) allows users to quit.

For much of 2023 and 2024, OpenAI’s GPT-4 maintained a defensible lead in reasoning quality and multimodal performance, which kept switching costs high for enterprises and individual users. But that moat narrowed considerably.

Rival models are reaching parity in core zones of competence, letting users switch. Market data reflects this. ChatGPT’s share of the Generative AI chatbot market declined from over 86% in 2024 to about 64.5% by early 2026 (bit.ly/4coOLwQ). While this contraction does not portend collapse, given OpenAI’s brand equity and reach, it signals the rise of a multi-polar AI economy.

Competitors have been swift to capitalize on market fluidity. Google’s Gemini is the most dramatic beneficiary. Its market share expanded from 5.7% to 21.5% within a year and it claimed 750 million monthly active users by February (bit.ly/4kx9Ua5). This surge is closely tied to the rollout of Gemini 3 and its integration across Android and Workspace, which has embedded it into daily workflows; Google Search has begun using Gemini too. By leveraging its existing user base, Google has converted ecosystem depth into rapid scale.

Other entrants have pursued differentiation rather than ubiquity. DeepSeek, for example, has disrupted pricing by offering reasoning-intensive models such as R1 at a fraction of prevailing costs, amassing over 96 million monthly active users by mid-2025 (bit.ly/4qwpGDy). Perplexity AI has consolidated a niche as a research-oriented engine, sustaining 45 million monthly active users. Together, the trends indicate that OpenAI’s lost share is not merely about ideological migration, but also the appeal of specialized value propositions made by rivals—cost efficiency, research precision or ecosystem integration.

Also, OpenAI is not the only one being ‘quit’ too. Platform churn is now common across the sector. Even Gemini’s ascent is tempered by a distinct form of attrition, as some users revert to traditional search tools or explore local-first AI models amid concerns over latency and the expanding footprint of AI-generated ‘overviews.’ So quitting has evolved into a recurring strategic recalibration rather than a singular act of protest.

Platform hopping does little to mitigate the structural threat these systems pose to Software-as-a-Service (SaaS) providers and the IT service firms that implement SaaS. As models evolve into active work participants, they begin to subsume functionalities that previously justified discrete software layers.

The recent market reaction to Anthropic’s Claude CoWrk puts this in sharp relief. Atlassian, producer of widely used team productivity and development tools such as Jira, has experienced a roughly 20% decline in market value since that launch. Snowflake, a data cloud platform, saw a decline of roughly 13%, reflecting investor concerns that AI systems reduce the frequency of direct data queries.

Rather than use multiple discrete interactions with a data cloud, CoWrk promises to retrieve information, analyze it and deliver finished outputs—tables, charts or structured summaries—in a single context-aware sequence. By retaining memory across tasks and re-using prior results, such systems diminish the transactional intensity upon which many current revenue models depend.

The stocks of various IT service providers, including major Indian firms, have seen a big sell-off in what has been dubbed the ‘Claude Crash.’ Such a sudden market response usually represents a credible power shift.

What appears to have unsettled investors most was not Claude’s raw capability, but its architectural trajectory. Anthropic signalled an evolution from tools that merely assist discrete human tasks to systems capable of retention and acting with a degree of autonomy over time. The rise of AI that does not simply answer, but remembers and executes tasks poses a more profound challenge than incremental improvements in reasoning accuracy. It threatens to compress entire layers of software and services into a unified intelligent substrate.

Within this transformation, the QuitGPT movement is best seen not as an isolated boycott, but as a manifestation of a better contested AI market. Users are discovering that loyalty is negotiable, switching costs are falling and alternatives are viable. Investors are recognizing that as AI systems gain persistence and agency, the value chains built atop traditional software may erode. Both consumers and capital are recalibrating their assumptions in real-time this year.

The author is co-founder of Siana Capital, a venture fund manager.