Domyn, an AI unicorn, is launching a new frontier model called 'Dom in Large' on Microsoft Foundry, specifically designed for regulated industries.
The company's core strategy is to provide clients with full ownership of their AI models, enabling them to maintain data sovereignty, ensure compliance, and customize the AI with their own domain expertise.
To overcome the global AI compute shortage, Domyn is partnering with NVIDIA and G42 to build its own supercomputer in Europe, highlighting the geopolitical importance of compute resources.
CEO Uljan Sharka predicts major shifts in AI, including the unification of training and inference, the rise of distributed compute disrupting centralized power, and another 'ChatGPT moment' arriving next year.
12 quotes
Concerns Raised
The global shortage of AI compute is a significant bottleneck for growth and training frontier models.
Low adoption rates of AI in regulated industries due to a mismatch between technology capabilities and enterprise needs for trust and control.
The current pre-training paradigm for AI models is unsustainable and creates static, non-learning systems.
Opportunities Identified
Providing AI model ownership to regulated industries is a major 'white space' market opportunity.
Building proprietary supercomputer infrastructure to secure a competitive advantage in the face of compute scarcity.
The development of smaller, domain-specific models that can outperform larger, general models on specific tasks.
The rise of distributed compute could level the playing field and disrupt the dominance of current market leaders.