How do you get to artificial general intelligence? Think more easily

In 2025, entrepreneurs will unleash a flood of AI-powered apps. Finally, generative AI is living up to the hype with a new generation of affordable consumer and business apps. This is not the general opinion today. OpenAI, Google and xAI are in an arms race to train the most powerful Large Language Model (LLM) in the quest for artificial general intelligence, known as AGI, and their gladiatorial battle is dominating the mindshare and revenue share of the fledgling Gen AI ecosystem .

For example, Elon Musk raised $6 billion to launch newcomer xAI and purchased 100,000 Nvidia H100 GPUs, the expensive chips used to process AI. Training its model Grok cost more than $3 billion. At these prices only techno tycoons can afford to build these huge LLMs.

The incredible spending by companies like OpenAI, Google, and xAI has created a one-sided ecosystem that is heavy at the bottom and light at the top. The LLMs trained by these massive GPU farms are also typically very expensive for inference, the process of entering a prompt and generating a response from large language models that is embedded into each app using AI. It’s like everyone has a 5G smartphone, but data usage is too expensive for anyone to watch a TikTok video or browse social media. As a result, excellent LLMs with high inference costs have made the proliferation of killer apps prohibitive.

This one-sided ecosystem of ultra-rich tech moguls fighting each other has enriched Nvidia while forcing application developers into the bind of either using a low-cost and underperforming model that will disappoint users or facing exorbitant inference costs and risk-taking bankruptcy.

In 2025, a new approach will emerge that can change all that. This will draw on what we have learned from previous technology revolutions, such as the PC era of Intel and Windows or the mobile era of Qualcomm and Android, where Moore’s Law improved PCs and apps and lower bandwidth costs improved mobile phones and apps year after year year.

But what about the high inference costs? A new law on AI inference is just around the corner. Inference costs have fallen by a factor of 10 per year, driven by new AI algorithms, inference technologies, and better chips at lower prices.

As a reference point, if a third-party developer were to use OpenAI’s top models to build AI search, the cost would be about $10 per query in May 2023, while Google’s non-Gen AI search is $0.01 Dollars would cost 1,000 times the difference. But by May 2024, the price of OpenAI’s top model dropped to about $1 per query. With this unprecedented 10x per year price drop, application developers will be able to use increasingly higher quality and lower cost models, which will lead to the proliferation of AI apps in the next two years.