The instability of Stability AI’s open-source path

Stability AI faces severe financial instability after projecting a $153 million loss in 2023 and experiencing a mass exodus of core research talent. High computing costs and legal battles with Getty Images threaten the company’s ability to sustain…

The instability of Stability AI's open-source path

Financial hemorrhage and leadership exits

Stability AI projected a $153 million loss in 2023 and generated only $11 million in revenue. The company spends $8 million every month. In November, revenue reached $3 million, while February saw $5.4 million. This deficit led the company to seek a buyer, though no deal emerged. Mostaque resigned in March 2024 after investors grew discontented with the company’s direction. This departure happened alongside the exits of chief operating officer Shan Shan Wong and chief technology officer Christian Laforte, who now serve as co-CEOs. The leadership vacuum grew even deeper when the core research team, including Robin Rombach, Andreas Blattmann, and Dominik Lorenz, left the firm. These researchers provided the technical foundation for the Stable Diffusion models. I view this mass exodus of talent as the primary reason the company struggles to maintain its technical edge against competitors like Midjourney. Stability AI faces a massive $99 million annual bill for processing power from Amazon Web Services, Google, and CoreWeave, a cost that frequently led the company to miss scheduled payments to its providers. By October 2023, the company held less than $4 million in its bank accounts. To reduce expenses, the company laid off about 10% of its workforce. I find the attempt to resell leased GPU capacity to cover revenue shortfalls a desperate move that undermines the company’s credibility.

Legal risks and model competition

Legal threats target the company’s foundational data. Getty Images accused Stability AI of copying over 12 million images without permission. This lawsuit, along with claims from a group of artists, creates unpredictable liabilities for the firm. You already know that open-source models provide more flexibility for developers, but Stability AI lacks the cash to keep those models updated. The company failed to secure commercial deals with Samsung, Snap, Canva, and several governments.

Feature Stable Diffusion 3.5 Midjourney v7 ChatGPT Images 2.6
Pricing Free for local use Monthly subscription ChatGPT Plus
Control High customization Artistic presets Conversational
Access Open weights Closed service Closed service

Stability Diffusion 3.5 relies on open weights, but it requires more technical tuning than commercial rivals to match high-end quality. Midjourney v7 provides polished, high-contrast images that require minimal prompting. Midjourney users pay for monthly plans such as the $30 Standard tier or the $60 Pro tier. ChatGPT Images 2.6, released in November 2026, focuses on conversational refinement. The team released Stable Audio 2.0 this month, which creates music files from text prompts. Last month, the team released Stable Video 3D to produce three-dimensional objects from images. While Halsion holds the title for the best overall quality in 2026, Stability’s community models provide more style diversity. Will the new leadership stabilize the company before the cash runs out?

The enterprise struggle

The company sold Clipdrop to the startup Jasper in February 2024, less than a year after its acquisition. This sale attempted to boost liquidity after the company burned through much of its $101 million funding from 2022. Stability AI secured $50 million through a convertible note from Intel in November 2023 to manage its cash flow. This funding followed a period where Mostaque faced accusations of exaggerating achievements and making misleading claims about partnerships. The departing researchers turned Latent Diffusion into Stable Diffusion. Mostaque used the $101 million from Coatue and Lightspeed Venture Partners to hire these researchers. The company continues to face competition from specialized tools like Ideogram 2.0, which focuses on text and logos. I think the company’s open-source focus provides its only remaining advantage, but its financial state makes that advantage difficult to sustain. The company currently has 26 open positions on its career site, whereas OpenAI has 36. Mostaque also suggested that the company could grow through collaboration with organizations needing foundation models as a service. The company’s management previously suggested it could manage services by reselling computing power, but this drew criticism. The company also earns revenue through its API and the commercial browser version, Dreamstudio. These streams attempted to offset the massive costs of training large-scale models. Mostaque aimed to build a model for national infrastructures.

airtrain.ai
airtrain.ai

The airtrain.ai newsroom covers AI research, models and the tools built on them.

More on this topic

Stay ahead of AI

Get the week's most important AI stories delivered to your inbox every Monday.

No spam. Unsubscribe anytime.

More Stories