The Future of AI: Applications Over Models

In recent years, the tech industry has witnessed an unprecedented surge in investments and innovations in artificial intelligence (AI). While foundational models have garnered significant attention and funding, the real potential of AI lies in the application layer. This shift in focus is driven by the belief that future AI unicorns will emerge from the development of advanced applications rather than the creation of models.

Sequoia Capital partner Pat Grady highlighted this trend during a gathering in San Francisco. The firm has invested heavily in companies like OpenAI, Safe Superintelligence, and xAI, which build expensive foundational models. However, Grady emphasized that Sequoia has invested even more at the application layer, despite the short-term revenue challenges faced by these applications.

The tech industry is increasingly viewing AI as a long-term play, with venture capitalists and leading platform providers investing heavily in large language models (LLMs). These models are computationally and capital-intensive to build, as explained by Charlotte Dunlap, GlobalData research director. Big Tech companies are pouring billions of dollars into training and inference of their AI models, with training costs reaching unprecedented levels. For instance, Google's Gemini Ultra and OpenAI's GPT-4 have incurred training costs of $191 million and $78 million, respectively.

Meta's decision to release its open-source model Llama has further complicated the revenue generation for proprietary models. OpenAI, for example, reported a $5 billion loss in 2024. As a result, investors are now turning their attention to the application layer of generative AI (GenAI) innovation. Applications are becoming more complex, utilizing multiple LLMs to power advanced apps and automatically route prompts to the most appropriate models.

The next-generation AI applications are focusing on data integration, with features like text to SQL and new forms of data access to improve business processes and enhance user experience (UX) and customer experience (CX). According to GlobalData's Artificial Intelligence Executive Briefing, the total AI market is projected to be worth $1,037 billion by 2030, with the specialized AI applications market accounting for $512 billion.

Snap Inc. is an example of a company that has heavily invested in AI to drive its applications. The company has developed augmented reality applications for its SnapChat platform and invested in computer vision engineering for future hardware modalities. With a growing consumer appetite for greater platform immersion and advances in computer vision engineering, Snap is positioned to become one of the next-generation Big Tech companies.

Qi Pan, who leads Snap's computer vision engineering, explains that the explosion in technology has significantly improved the capabilities of computer vision. Snap's lens creators have built millions of lenses, viewed by trillions of users, showcasing the phenomenal impact of creator-generated content.

The business utility of AI lies in the application layer, as noted by Ignasi Barri, global head of data and AI at GFT. Companies need to identify the right use cases to leverage LLMs or Large Vision Models (LVMs) to create business opportunities and achieve a return on investment. Specialization and agility allow smaller players to target niche markets and solve specific problems with AI-driven solutions.

Vijay Guntur, CTO of HCLTech, emphasizes the importance of developing AI applications tailored to specific industries like healthcare and pharma. AI can predict the success of drug discovery processes, reducing risks and increasing the number of new drugs launched annually. The AI boom is expected to spawn giants, with companies seeing a return on investment by developing their own applications.

In conclusion, the future of AI is bright, with the application layer poised to drive significant business utility and innovation. As companies continue to invest in AI applications, we can expect transformative value propositions and the emergence of new industry giants.

Cited: https://www.verdict.co.uk/future-ai-unicorns-will-emerge-from-the-application-layer/?cf-view