I remember more than 25 years ago, during the .com boom, it was an exhilarating time to be alive. There were so many .com startups. Everything online was new and free. There were tons of free online services.
I remember signing up for my first free email account from the now-defunct and unheard-of free email service from America. Back then, it was a big deal. It was a sign of prestige to have an email address. It made you feel important because you will be contactable by anyone in the world. However, the problem was that almost everyone got their email accounts from their ISP, school, university or work. If you changed your ISP, graduated from school or switched jobs, your email account will be deleted. So, for most people, their email addresses could not be their permanent identity. My free email account from the free online service was the beginning of what I thought to be a permanent online identity. I would have the same email address regardless of which school or university I attended, which ISP I connected from and which jobs I switched to. Then I found out that there were so many free email services on offer. Since they were all free, I signed up for as many as I could, like a child gobbling up as much candy as he could.
Back then, Google was a cool and intriguing start-up. They were the good guys by providing so much free stuff for the world. The first piece of free stuff I got from Google was a Gmail account. Then I signed up for many of their other free services like Reader, Picassa, Blogger, and so on. And then there were other free stuff from other start-ups. There were even a couple of free Web hosting services. I signed up for them to host the websites for my first book and affiliate website. Auction sites were also free back then, and I used them to sell my second-hand stuff.
Back then, as a budding entrepreneur, it felt that making money online was going to be easy even though the reality proved to be different. Even if you did not want to be an entrepreneur, there were plentiful IT jobs around. As an entrepreneur told me years later, the tech industry was so short of tech workers that it was said they hired anyone who could breathe. Of course, that saying was a hyperbole, but you get the idea.
The prevailing idea back then was that .com companies should create products (with investors’ money), give them away for free to gain massive market share, and then figure out a way to monetise them later. Sustainable profit was a secondary concern.
Then the .com bubble burst spectacularly. Then it was followed by the GFC. Then around 10 to 15 years ago, there was a secondary mini .com boom that went nowhere. Facebook (now Meta) arose from that secondary boom.
Today, Google and Amazon were some of the few that survived the bursting bubble, thrived and became a global empire.
.AI boom
Today, history is repeating itself. Instead of the .com startups of the past, we now have the .ai startups.
Once again, I am seeing a lot of .ai companies providing free or low-cost Artificial Intelligence (AI) services. I am seeing AI services like text prompts to videos, and images, voice cloning, avatar face cloning, AI influencer creator tools, and so on. There are even AI services that can construct entire websites via text prompts! Microsoft is also letting their people use CoPilot for free. Social media like Facebook, Instagram and LinkedIn are incorporating AI into their platforms. Canva and Adobe are including AI in their design tools. You can even use AI to write code for you.
I am seeing parallels with the .com boom. Back then, I felt that making money online was going to be easy because of the proliferation of huge numbers of free platforms and services that can be used by entrepreneurs.
Today, it feels déjà vu again. Once again, all these free and low-cost AI services promise to unlock massive productivity multipliers that can lower costs, and eliminate the need to hire humans by automating large parts of your business processes.
The same alluring promise of easy money has come back. AI is going to cut your business costs and make running your business so much easier than before that your profit is going to soar. Those businesses that do not learn how to leverage AI are going to be left behind, become dinosaurs and go out of business because they will be unprofitable because of relentless and ruthless competition from businesses that know how to use AI.
Even if you do not want to be an entrepreneur, you have to worry that if you do not know how to use AI, you may not keep your job.
So now, there is a mad rush to learn how to use AI. Interest in AI has soared.
During the gold boom, the people who made money were the ones who sold the shovels to the gold diggers. So, I suspect that in this .ai boom, the ones who will be making money will be the ones teaching people how to use AI.
How the allure will fade
The .com bubble burst because the startups did not pay attention to the financial basics. They did not work out how to make long-term sustainable business profits.
In the same way, I suspect a lot of .ai startups are going to face the same fate. The easy part is to burn investors’ money to build the product. The hard part is to make the business sustainable by making sustainable profits in the long run.
Cost of providing AI services
Make no mistake, the cost of providing AI services is not cheap.
It costs a lot of money to invest in building large data centres to hold the huge amount of data needed to train the AI. Training AI requires a lot of money to buy lots of computer chips to do the data crunching. Computer chips require a lot of money to pay for the electricity costs to power them and water to cool them. As this Washington Post article reported,
The immense cost of training AI algorithms — which requires running mind-boggling amounts of data through warehouses of expensive and energy-hungry computer chips — means that even as companies like Microsoft, OpenAI and Google slowly begin charging for AI tools, they’re still spending billions to develop and run those tools.
Recently, Sam Altman from OpenAI asked for up to $7 trillion for an AI project.
Yes, it costs trillions of dollars!
In other words, given the astronomically high cost of providing AI services, how can all these free and low-cost AI services be profitable in the long run?
Big tech companies are pouring billions into AI projects and they are losing lots of money. So, two things are going to happen. Either the projects will fail, startups will go bust, or they are going to start charging a decent amount of money to make AI services sustainable financially.
Running AI locally
To give you an example of the amount of processing power of running AI, try downloading and running the Diffusion Bee app (if you have a Mac).
This app allows you to run generative AI locally on your Mac. You give the app a text or image prompt and it will generate or process an image for you. Each image generation takes about 30 seconds for an M1 Mac with 8 GB of memory. I have tried it and the resolution of the image it produced is quite low. It is so processor intensive that it will hog up the processing power of your Mac.
Running AI in the cloud?
Now, imagine running this generative AI in the cloud.
To make it useful, it needs to produce images of much higher resolution than what Diffusion Bee can do locally on your Mac. This will consume a lot more processing power.
Also, to be useful, it has to be fast too. Preferably 15 seconds or less. How much more processing power will that consume?
Now imagine, each person like you using this generative AI a gazillion times per day. To produce something useful, you need lots of iteration of experimentation until you come up with a text prompt that creates something that you want for your purpose. Generative AI is not going to produce something you want on the first try. Sometimes, you need several tries at least. Perhaps even more.
Now, imagine that you are trying to make generative AI produce a decent video with your prompt. Imagine running this generative AI through the cloud.
Next, imagine hundreds of millions of people (maybe even billions) running generative AI in the cloud. Think of how much energy all this computer processing is going to consume just to create amazing generative AI images and videos in the cloud for hundreds of millions of people banging away on their keyboards from home and office every day, 24/7.
Is this even sustainable in terms of global energy consumption?
I can foresee that there is going to be a need to ration generative AI services. The rationing will be done by charging a price for their services.
Will interest in AI be sustainable?
Let’s say these AI startups learn the lessons of the .com bust and start charging a decent amount of money for their service. What will happen?
You can bet that the current interest in AI will tumble. Yes, AI can be amazing. But if it costs too much to use them, businesses and people are going to scale back their use and choose wisely which ones to pay for. It may turn out that only a few Big Tech AI companies will be able to charge enough to provide good AI services to enough people to make it profitable in the long run.
My advice to you
Have fun experimenting and playing with AI while you still can while they are still cheap.
One day, you will have to choose wisely which AI service to pay for.