AI Secret Bubble 3 hidden problems that will bankrupt your business

AI’s Hidden Bubble: 3 Financial Challenges Threatening AI Companies

From our analysis, they are set to implode nearly every company that relies too heavily on AI in less than 2-3 years.

Anyone that’s been working or integrating AI into their business models has discovered a harsh reality: AI has two profitability issues.

Currently, nearly all of the foundational generative AI models and multi-modal models are sustained by constant rounds of investments from organizations riding the hype train.

I understand listening to the promises of folks like Sam Altman (CEO of ChatGPT’s OpenAI) who are appearing in every media format they can access.

Their goal is obviously to proclaim that “AGI (artificial general intelligence) is just around the corner… so invest now.”

Altman is seeking trillions of dollars because he and everyone who knows how this all works understands that AI has a massive profitability issue.

Why? There are three profit-pits as to why the AI foundational model companies are not profitable. This is why they will not achieve any semblance of commercially available AGI for the foreseeable future.


Money Pit #1: The AI Chip and GPU Shortage

There’s a reason why NVIDIA’s stocks are skyrocketing and the demand for their computer chips has exploded since generative AI became the hottest topic at the end of 2022.

AI models, especially multi-modal systems, require a hefty amount of computational power for both their training process and to run the AI models at scale.

Every company is trying to get their hands on the best computer chips and graphics cards, resulting in lower supply than demand.

This lack of AI-focused computer chips is slowing down the development process.

Once they become available, the cost will be astronomical for any business to invest in that kind of silicon. Most companies aiming to reach AGI will be unable to due to this limitation.

What does the AI chips problem mean for your business? It means that the costs for connecting to their API will increase.

Just as OPENAI’s ChatGPT API is currently expensive, even enterprise users are reconsidering its viability.

If these costs are passed onto their end-users per token request, churn rates or subscriber loss would spike considerably.


Money Pit #2: The High Cost of Ethical AI Training Data

Tech companies will soon need to license their training data from the actual copyright owners.

This is the main reason they initially claimed “Fair Use” (which is not fair use when commercialized).

The costs associated with ethically licensing that data will skyrocket, weakening the impressiveness of their models.

Training large-scale AI models incurs massive costs due to the need for substantial computational resources, data handling, and infrastructure.

There was a recent projection that the cost of training more sophisticated AI models will soon reach $10 billion, instead of the current hundreds of millions.

What does the training data problem mean for your business?


Training Data Transparency

It means that all companies making generative AI from these foundational models, or if your business has built its own generative AI model, need to consider this issue.

If you’re using “publicly available” data from the internet without compensation or consent, or using data from big tech companies that haven’t compensated or sought consent, your business could face legal challenges.

Each foundational AI model company has at least a handful of lawsuits from copyright holders, including Google, OpenAI, Anthropic, Microsoft, StabilityAI, and more.

These lawsuits started at the end of 2022 and generally take 2+ years to resolve.

Your favorite companies are integrating these AI foundational models’ APIs into their products and services, from Salesforce (likely ChatGPT API), to Canva (using StabilityAI), Grammarly (ChatGPT API), and more.

Once the hammer drops, and it will, the incoming legislation and ethical corrections in the market will strip your business of what makes your current AI model impressive.


Don’t Bet The House On AI Right Now

If you’re betting on this type of artificial intelligence to be the main value driver of your business, reconsider.

Once the law catches up and lawsuits mandate that companies strip their AI models of copyrighted and unconsented training data, it will drastically downgrade the capabilities of the AI models you’ve integrated into your products, services, and business models.

Ultimately, this could crash not only your main value driver but your business as a whole.

This is why AI CEOs like Sam Altman are trying to secure deals with publishers and film studios now, although they didn’t do this before.

They are preparing, but the data they will acquire from these deals with lower-tier publishers and film studios will pale in comparison to the billions of data they initially took off the internet.


Money Pit #3: The Massive Energy Costs of Running AI Models

Currently, the biggest cost for all AI foundational models is their energy costs to run their models.

All foundational models are spending $10 million per month in energy costs to run their AI models at scale.

It’s still not enough, which is why the majority of them are constantly limiting power consumption.

For those worried about Terminator Skynet or Matrix robot revolution, that’s not going to happen any time soon.

At least until they solve the lack of electricity and energy to run these models at scale, which is not just a money issue but a technology issue.

Some believe that proper commercial AGI will not be achieved until we introduce nuclear energy to run the AI foundational models or achieve sustainable energy strong enough to power all the AI data centers.

What does the AI energy cost issue mean for your business? It means your usage of AI will continue to be throttled for the sake of power consumption, and the energy costs will be included in the per-token request if you are connecting to the API of the AI companies.

When a company cannot achieve profitability, the cost will be levied on the user, making it more expensive over time.

If the AI company cuts corners, they may “foot the bill” by doing something with your business data that you may not have consented to.


The Reality of AGI Timelines and Hype

As a tech entrepreneur, my team and I have been working on AI software for the past 4 years and have been observing this emerging technology since it proliferated in 2022.

From our analysis, AGI is not around the corner without addressing these three problematic money pits.

The only reason folks are saying this, and in some cases, even faking demonstration videos such as Sora’s Balloon Man video or Google’s faked multimodal demo, is that they are putting out proofs of concept and stringing us all along to make specific investors believe that AGI is coming with the next investment… and the next investment… and the next investment… so they can continue to play with their toys in the lab.

Share this post