In March, we launched this series, and I must say, I’m continually surprised by how many intelligent people still don’t realize that we are just at the dawn of the Age of Artificial Intelligence. Looking back, there have been five distinct technological ages since the advent of the personal computer (PC): the rise of the Internet, mobile computing, the cloud, and now Artificial Intelligence (AI). Each wave of innovation has built upon the last.
Recently, some investment banks and others—who, quite frankly, missed the initial AI shift—have started drawing comparisons between AI and the “internet bubble.” While there are some similarities, the differences are far more significant. It’s like saying that all businesses are the same because they use double-entry accounting, have human resources departments, or generate similar revenues. No, AI is fundamentally different.
Sure, billions of dollars in investments are pouring in from massive companies, and the cost of using AI Large Language Models (LLMs) has plummeted, with some costs dropping by as much as 99%. See the graph below, borrowed from Andreessen Horowitz, a leading Silicon Valley firm, which illustrates this dramatic decline.
There are a lot of data centers already in operation, with many more planned. This is a positive development for society, as it makes AI cheaper and more accessible. However, it will have far-reaching implications for productivity and employment, both of which will be profoundly disrupted.
LLM Tokens – The Cost Structure
Let’s take a moment to explain the concept of tokens in relation to LLMs. This is a different use of the word “tokens” than what we typically associate with blockchain or cryptocurrency.
Large Language Models (LLMs) like ChatGPT or Claude charge users based on tokens, which are essentially chunks of text. A token can be as short as a single character (like “a”) or as long as a full word (like “apple”). On average, 100 tokens are roughly equivalent to 75 words in English. Both your input (what you type) and the model’s output (the reply) consume tokens, and you are billed according to the total usage.
LLM providers often offer various pricing models, with more powerful models typically costing more per token. Some services, like ChatGPT Plus, offer subscriptions, while others use Application Programming Interface (API) pricing for developers and businesses, with costs decreasing as usage increases. This is particularly relevant for companies developing Agent AI or Agentic AI tools, as they are billed based on this token structure.
For example, generating a long email reply could use 300–500 tokens. If a model charges $0.002 per 1,000 tokens, that single reply could cost just a fraction of a cent.
I will use available data (which has likely decreased since) to illustrate the relative costs. The latest version of GPT-4o, which powers the top-rated ChatGPT, now costs $2.50/$10 per million input/output tokens— a significant reduction from the previous rate of $5/$15. GPT-4o Mini costs just $0.15/$0.075 per million tokens. Llama 3.1 405B is available for $2.70 per million tokens, which is about 66% less than Azure. Gemini 1.5 Flash costs $0.15/$0.60 per million tokens, while DeepSeek v2 costs $0.14/$0.28 per million tokens.
As you can see, there’s a pricing war in the AI space, which is great for companies using LLMs in their products, for businesses seeking to become more competitive, and ultimately for consumers. However, this also has significant consequences for workers in affected industries.
You can use a “Token Calculator” to estimate how many tokens you’ll use. For example, when I typed “this is my name, Noel,” the calculator showed that I used 6 tokens, 17 characters, and 21 total spaces.
The Internet “Bubble” vs. Today’s AI Development
The key difference between the internet bubble and today’s AI development lies in the support behind LLM businesses. Unlike the early days of the internet bubble, where the startup ecosystem was the primary driver, today’s LLM development is backed by the largest and most capitalized companies in the world, such as Microsoft, Oracle, Google, Meta, SoftBank, Nvidia, xAI, and OpenAI.
For many of these companies, AI is just one small part of their overall business, but they are still heavily invested in its success. This is crucial because the cost of developing LLMs is so high that only the biggest players could initially afford to get involved. However, that is changing. As the cost to build smaller language models has fallen by about 90%, more players are entering the space.
However, we believe there is a bubble in the AI Agents and Agentic AI space. These companies often use LLMs as the core of their service, but they lack intellectual property that is patentable or even proprietary. This is why, as we’ve mentioned before, the true value lies in the data. The key question is: How big is your moat? How accessible or unique is your data?
The World Economic Forum (WEF) Future Employment Report 2025
The World Economic Forum’s Future of Jobs 2025 report, an impressive 290-page document, should serve as a wake-up call for everyone. We’ll delve deeper into this report in the next blog, but the bottom line is that we’re probably not adapting fast enough to the rapid changes that AI is already bringing to our world.
How AI Will Change The Consumer – Forever
We’ve often discussed the benefits for consumers in having their entire electronic health record (EHR) in one place. Imagine a Universal Health Record (UHR)™ that’s integrated with carefully curated AI (not LLMs) to inform consumers about their health status, help them communicate with healthcare providers, and even discover potential clinical trials. This service could also empower consumers to challenge insurance companies when they deny care or medication.
Our goal has always been to empower consumers and advance healthcare science, especially in a system that is increasingly complex and, at times, adversarial to consumers.
About HealthScoreAI ™
Healthcare is at a tipping point, and HealthScoreAI is positioning to revolutionize the industry by giving consumers control over their health data and unlocking its immense value. U.S. healthcare annual spending has exceeded $5 trillion with little improvement in outcomes. Despite advances, technology has failed to reduce costs or improve care. Meanwhile, 3,000 exabytes of consumer health data remain trapped in fragmented USA systems of 500 EHRs, leaving consumers and doctors without a complete picture of care.
HealthScoreAI seeks to provide a unique solution, acting as a data surrogate for consumers and offering an unbiased holistic view of their health. Giving Consumers tools to respond to denial of care by insurers, we aim to bridge gaps in healthcare access and outcomes. By monetizing de-identified data, HealthScoreAI seeks to share revenue with consumers, potentially creating a new $100 billion market value opportunity. With near-universal EHR adoption in the USA, and advances in technology, now is the perfect time to capitalize on the data available, practical use of AI and the empowering of consumers, in particular the 13,000 tech savvy baby boomers turning 65 every single day and entering the Medicare system for the first time. Our team, with deep healthcare and tech expertise, holds U.S. patents and a proven track record of scaling companies and leading them to IPO.
Noel J. Guillama-Alvarez
https://www.linkedin.com/in/nguillama/
+1-561-904-9477, Ext 355