AI Startup Frenzy: Billions Raised, Little to Show

AI Startup Frenzy: Billions Raised, Little to Show

The artificial intelligence sector is experiencing an unprecedented surge in venture capital funding, with startups like Thinking Machines attracting billions despite lacking a tangible product. This phenomenon highlights the intense hype surrounding AI and the willingness of investors to bet heavily on perceived future potential, often based on the pedigree of founding teams rather than demonstrable market traction.

Thinking Machines: A Case Study in Hype

Thinking Machines, founded in February 2025 by former OpenAI executive Mira Murati, exemplifies the current state of AI startup investment. Murati, who played a significant role in the high-profile leadership changes at OpenAI in 2023, assembled a formidable founding team composed of 30 AI researchers and engineers, many of whom were her former colleagues from OpenAI. Key figures included John Schulman, a co-founder of OpenAI, as Chief Scientist, and Barrett Zof, formerly VP of Research at OpenAI, as CTO.

The company’s initial public statements outlined ambitious goals: assisting users in adapting AI systems, developing foundational AI capabilities, and fostering scientific understanding within the field. These declarations, coupled with the strong credentials of its founding team, set the stage for a massive fundraising effort. In July 2025, Thinking Machines secured a staggering $2 billion in its seed funding round, valuing the company at $12 billion. This marked the largest seed funding round in history at the time, with investors including prominent venture capital firms like Andreessen Horowitz and Lightspeed Venture Partners, as well as tech giants such as Nvidia, Cisco, and AMD.

Adding an unusual dimension to the funding, the government of Albania contributed $10 million. This investment appears to be rooted in Murati’s ethnic heritage and birthplace, a move described as an “eccentric way to manage public finances” given the company’s base in San Francisco and lack of operations in Albania.

Product Launch and Market Reception

Despite the enormous capital infusion and the promise of building “models at the frontier of capabilities” and “multimodal AI,” Thinking Machines’ first product, named Tinker, launched in October 2025, fell far short of expectations. Tinker is a tool designed for developers to fine-tune open-source large language models (LLMs). Fine-tuning involves further training a pre-trained LLM on a specialized dataset to improve its performance on specific tasks or domains.

While fine-tuning is a valuable process, Tinker’s utility is significantly limited by its exclusive support for open-weight models, such as Meta’s Llama and DeepSeek. It cannot be used with leading proprietary models like OpenAI’s GPT series, Anthropic’s Claude, or Google’s Gemini. This restriction sharply curtails its market potential, as the most advanced and widely adopted AI models are currently closed-source, with their developers unwilling to share model weights. Data from Similar Web indicated that in late 2025, ChatGPT commanded 68% of AI tool web traffic outside China, with Gemini at 18%, leaving open-weight models with a much smaller share.

Tinker’s commercial traction has been minimal since its launch. The company’s website features only four customer testimonials, primarily from academic and research institutions like UC Berkeley, Princeton University, and Stanford. There is little evidence of adoption by large corporations for commercial use cases, suggesting negligible revenue generation for the product.

Red Flags and Team Departures

Further raising concerns, visual evidence presented during the Tinker launch suggested the potential fabrication of marketing materials. A billboard image advertising the Tinker API, shared by the company, exhibited anomalies – a lack of photographic grain and unnatural sharpness – leading to speculation that it was digitally manipulated to falsely convey a more established market presence.

The situation at Thinking Machines began to unravel significantly in late 2025 and early 2026. In November 2025, reports emerged that the company was seeking to raise additional capital at a $50 billion valuation, a five-fold increase from its previous valuation, a move considered highly ambitious given the underwhelming product and limited traction. This funding round has yet to close.

More critically, the company’s core team began to disintegrate. In October 2025, Meta successfully recruited Andrew Tolk, a co-founder of Thinking Machines and former OpenAI researcher, after a previous lucrative offer. Subsequently, on January 14, 2026, Mira Murati announced the departure of CTO Barrett Zof, citing an undisclosed affair and “performance, conduct, and trust issues.” Strikingly, on the same day, OpenAI announced Zof’s return to the company, along with two other Thinking Machines co-founders, Luke Mets and Samuel Shonholtz.

With four out of seven founding team members having departed, the credibility and future prospects of Thinking Machines are severely compromised. The mass exodus suggests a potential loss of confidence in the company’s direction and leadership, exacerbated by competitive offers from tech giants like Meta and OpenAI, who are known to offer substantial financial incentives, including signing bonuses reportedly as high as $100 million for top AI talent.

The Wider AI Investment Landscape

The case of Thinking Machines, alongside that of Safe Super Intelligence – another OpenAI alumni startup founded by Ilia Sutskever, which raised $3 billion with no product after over a year of operation – underscores the irrational exuberance in the AI sector. Venture capital firms, particularly those like Andreessen Horowitz, which raised a dedicated $20 billion fund for AI investments in 2025, are pouring vast sums into speculative ventures. This “AI bubble” appears to be funding even the most “absurd ideas,” driven by intense market hype and the fear of missing out on the next transformative technology.

Market Impact and Investor Considerations

The extreme valuations and rapid fundraising seen in the AI sector, detached from traditional metrics of revenue and product-market fit, present significant risks for investors. While the potential of AI is undeniable, the current environment suggests a disconnect between perceived value and fundamental performance. For investors, this necessitates a cautious approach, emphasizing due diligence beyond team pedigree and hype. Understanding the specific technological niche, competitive landscape, and the realistic path to commercialization for AI startups is crucial. The rapid disintegration of founding teams and the underwhelming product launches seen with Thinking Machines serve as stark warnings about the volatility and speculative nature of the current AI investment climate.

Source: The Absurd Case of Thinking Machines (YouTube)