A valuation of $1 billion supported Character.ai when it raised $150 million in 2023, yet last month the chatbot platform filed for bankruptcy just five years after launching. The business unravelled despite high-profile backing, demonstrating how quickly capital disappears when software models fail to generate sustainable revenue. According to industry media, Google effectively acquired the founding team and licensed the technology of Character.ai, leaving the company to handle operating expenses on its own.
That corporate collapse is no isolated outlier in an overhyped sector. The market has ceased tolerating indefinite burn rates, as unsustainable infrastructure overhead and missing end-user demand erode corporate budgets. The central issue confronting business leaders is whether software organizations can sustain ongoing inference and operational expenses against deteriorating customer retention and shrinking margins.
The Anatomy of Corporate Abandonment
Scale alone offers no protection against the structural economics of modern artificial intelligence. According to the assessment of S&P Global Market Intelligence, about 42% of AI initiatives are ultimately abandoned by their corporate parents. The analytical firm attributes these shutdowns to insufficient funding, technical challenges, competition, difficulty scaling, and weak user demand.
"According to S&P Global Market Intelligence, about 42% of AI initiatives are ultimately abandoned by their corporate parents."
The rate of discontinued experiments reveals that nearly half of enterprise investments end in technical or economic dead ends before showing a return. While enterprise leaders routinely anticipate friction when deploying new systems, the scale of internal write-offs indicates that models frequently fail the basic requirements of enterprise unit economics. Why are foundational model creators and hyperscalers also shuttering live products?
Because the underlying operating metrics fail to support standalone applications once initial interest subsides. Even OpenAI has stumbled repeatedly when attempting to spin out independent consumer surfaces. Its video-sharing platform, Sora, struggled with high operating costs and user-retention challenges before shutting down in March 2026. Standalone ventures have faced a similar fate: ChatGPT Atlas, an AI-powered web browser, lasted less than a year before being discontinued on August 9, with its useful features absorbed into ChatGPT.
Big Tech Missteps and Infrastructure Dead Ends
OpenAI previously encountered friction with product consolidation as well. Standalone agent software like Operator suffered product consolidation, while image generator DALL-E saw its role as a dedicated destination reduced as OpenAI embedded generation features straight into ChatGPT.
Engineering friction has proven just as costly for hardware giants trying to market native intelligence. Apple highlighted an upgraded Siri when presenting Apple Intelligence in 2024, promising deep context awareness across applications. However, Apple repeatedly pushed back the rollout because of engineering issues and bugs.
This delay eventually contributed to a $250 million settlement over claims regarding how Apple marketed the AI capabilities of the iPhone 16.
The consumer hardware maker finally launched Siri AI inside the iOS 27 beta this past July, beginning an English-language rollout this month. For smaller market entrants, the pressure from platform vendors leaves virtually no runway. Relay, an AI-powered workflow automation tool built as an alternative to Zapier, shut down entirely in August 2026 after five years of operation as Google and OpenAI embedded comparable automated agents directly into baseline enterprise suites.
Liquidation and Structural Viability
Some argue that high failure rates simply reflect normal disruption during the initial phases of a technological transition, serving as a necessary precursor to a profitable software market. Yet the operational record indicates that technical achievements do not protect balance sheets from immediate compliance, safety, and infrastructure bills. Privacy-focused assistant Ollie launched in June 2026 as one of the first mainstream apps to achieve the SOC 2 compliance standard, yet the application announced its shutdown.
Corporate distress cases underline the reality that intellectual property values compress rapidly once cash reserves evaporate. In the pre-bankruptcy phase in September, Disney funneled $5.5 million alongside a refundable $2 million deposit to the startup, then hired its CEO and most engineers. Disney subsequently insisted that the transaction proceed through Chapter 11 proceedings because of the startup’s evaporating cash reserves and severe litigation exposure, ultimately offering an opening bid of $20 million to license intellectual property from a business once valued in the billions.
According to industry sources, the buyers insisted on structuring the transaction through bankruptcy proceedings due to high regulatory risks and the developer's liquidity deficit.
Will Disney close its $20 million acquisition of Character.ai assets by December, or will ongoing litigation and regulatory headwinds leave the venture with no solvent buyers?
Why it matters
A $1 billion valuation vanishes instantly once crushing infrastructure expenses hit vanishing user retention and an empty bank account. Standalone interfaces fail to justify high inference costs on their own, prompting even OpenAI to dismantle tools like ChatGPT Atlas and fold features back into a single hub. The failure of 42% of corporate AI projects comes down to broken unit economics rather than missing algorithms.
Our stories: Disney Acquires Character.AI Team via Bankruptcy Chapter 11 (October 20).