“The cost of using certain levels of AI capabilities drops by about 10 times every 12 months,” Openai CEO said in his 15-page proposal to the White House, highlighting the urgency to maintain the American edge in the AI race. This is a bold statement highlighting how quickly the economics of AI is changing, and what it means for the future of companies like Openai.
Openai’s leadership in AI is clear. The GPT-4O model sets the standard of quality and flexibility. But the same advancement that has driven its growth is rebuilding the competitive environment. Rivals like Google, Anthropic and Open-Source challengers are closing the quality gap while offering low prices, Opennai faces a difficult truth. That business model is under pressure.
The changing cost dynamics of AI
A few years ago, running AI models was not cheap. The GPT-3 costs around $60 per million, so only deep businesses can access them. Fast forward to 2024 and things look very different. Openai’s token costs fell 150 times between the launch of the GPT-4 in early 2023 and the debut of the GPT-4 in mid-2024. As Openai said, “Moore’s law predicted that the number of transistors in a microchip would double about every two years. The reduction in the cost of using AI is even more dramatic.”
“The amount of calendar time required to improve AI models continues to decrease. AI models are catching up to human intelligence at an increasing rate. The typical time it takes for a computer to defeat a human on a particular benchmark has decreased from five years after the benchmark was introduced, from five to two years. There is no reason for these advances to halt in the near future.”
This significant cost is perfect for expanding access, but it poses new challenges. Openai must find ways to keep its lead while making profits in a market where prices are falling apart.
Why Openai’s benefits are reduced
The GPT-4o remains the lead in reasoning, subtle responses and overall usability. However, its edge is shrinking for several reasons:
Price Pressure: Competitors like the Deepseek R1 are rolling out models that are approaching quality but far less expensive. Companies need to ask if the additional costs of the GPT-4o are really worth it. Easy Switching: AI is different from traditional cloud services where switching providers are painful. Developers can switch models with just a few lines of code. This makes it difficult to maintain loyalty. Open Source Surge: Open Source models like Mixtral have access to high quality AI for almost free, creating more pressure on Openai.
But Openai doesn’t just build models. Its strength lies in its mission to develop AI responsibly and make it easier to access. The challenge is to retain its advantages while facing fierce price competition.
Openai also has a big bet on infrastructure. The company is reportedly working on Stargate, a $500 billion data center project designed to make its AI business the future. This investment could ensure Openai has a stronger foundation to support more advanced models and retains the computing power to innovate faster than its competitors. In an industry where speed and scale are important, Stargate may need to open a long-term moat to protect its leadership.
Who really wins with AI?
Ironically, it’s not necessarily a model provider that makes a lot of money. The real winner is the “rapper.” It is a company that builds user-friendly products in addition to existing models.
Platforms like Concept AI, Confusing AI, and Quora Poe are thriving. They focus on creating smooth and intuitive experiences that help users get more from AI. Meanwhile, companies like Openai, Google and humanity are trapped in a cycle of heavy investment, cutting prices just to stay competitive.
For these “wrappers”, the model is merely an engine. The real value comes from how they package it, making it easier and more valuable for the user. This approach creates stronger customer loyalty and better long-term outlook.
How about Nvidia and Ai Chipmakers?
AI model providers don’t just feel the heat. Chip makers like Nvidia can also face trouble. As AI models become more efficient, fewer GPUs need to be used to run. This could damage Nvidia’s sales in the long run. Companies such as Broadcom and Marvell have already developed custom chips that will challenge Nvidia’s grip in the market.
That being said, Nvidia still benefits from the overall growth of its AI infrastructure. The pressure isn’t immediately, but the advantage of Nvidia’s advantage can slip as the AI model is more lean.
What’s next?
Open Source Options: The open source model continues to improve and offers low cost, solid performance. AI as a product: The real winner is a company that not only creates models, but also builds excellent products and experiences. Step Up for Big Tech: Google, Meta and others are closing the gap, and Openai’s leads are less safe. Price reductions continue. It forces model providers to rethink how profitability will remain in the hope that costs will continue to decline.
Big Questions from Openai
Openai still holds the crown, but it’s not easy to maintain. The market is changing rapidly, and price alone can be the determinant of many customers. The companies that appear on top are companies that build convenient and accessible products that solve real problems.
For example, Chinese AI startup DeepSeek has already proven that it is possible to fire major US AI companies. In January, Deepseek surpassed ChatGpt on the App Store, sending shockwaves through tech stocks. The talk around Deepseek began in December after the V3 model surpassed top US AI models, including Meta’s Llama 3.1, Openai’s GPT-4O, and Alibaba’s Qwen 2.5 on third-party benchmarks.
Openai may not be at the top for a long time. Some speculate that Openai is asking for a ban on Deepseek and other Chinese open source AI models. Deepseek operates with more than 2,000 Openai employees and just 160 employees, and that’s been a major advance.
Source link
