Here’s a sobering fact. The human created internet is dying quickly. Human web traffic is the minority at 42.6% Let that sink in. In 2026, 90% of new online content is AI-generated, 51% of US internet traffic is AI bots and 73% of readers can’t tell human vs AI work. We are feeding the machine at an incredible rate. Approximately 1 in 2 adults use AI chatbots; 60% under 30; that’s up 275% in 1 year. There’s currently 1,360 hyperscale datacenters built, another 1,500 proposed. The current human impact is enormous, the future impact, incalculable.
The AI hypderscale data center I used in the featured image is Amazon’s flagship Project Rainier AI data center campus located in New Carlisle, Indiana. It’s already built and operational at 6.5 million square feet with 32 data center buildings. It uses 2.2 to 2.25 gigawatts (GW) of electricity from the power grid. This is enough power to support approximately 1.5 million average homes and marks the facility as the largest single electrical customer in Indiana’s history. What does it do? Serves as a core engine for Anthropic to train its Claude AI models. That’sit, just training AI models.
Where does that data come from you might ask? It’s storage could hold the entirety of human knowledge. Well, that’s YOU, ME and THEIR data. All of it. Every last character you have ever typed that’s online, on your phone, on your computer, on heath/police/government/corporate data bases. All of it. All trained on.
That’s 1 hyperscale datacenter. ONE.
How does this affect the internet? It’s so titanic of an impact I had to read this twice myself.
Currently we have:
- Tens of millions of active, automated bot instances flooding the web
- 10+ million custom/unverified AI scraping scripts
- Over 1.2 million LLM models
- Over 1 million open-source model variants
- 100 to 200 prominent, closed-source enterprise models ( by companies like OpenAI, Anthropic, Google, Microsoft, DeepSeek and Mistra).
- 10 million to 30 million+ custom bots
- 50 flagship consumer AI platforms
- Tens of millions Custom & Agentic Bots
- Billions active users
That’s 1 AI Entity for every 6.96 (approximately 7) active websites. Dedicated, high-performance crawling architectures can scrape (steal) website pages at speeds of 100,000+ pages per second by bypassing heavy browser rendering and pulling raw text directly. For Total Motorcycle, we have 600,000 pages of high value human written content. That’s… 6 seconds, then POOF! Gone. Now multiply that by the over 1.2 million LLM models and not just OpenAI GPT, Anthropic Claude, Google, Meta Llama, DeepSeek, xAI Grok, Mistral AI and Alibaba Qwen and you get a solar system glimpse of the AI universe in front of us.
Impact: Active Human Content Websites: There are roughly 217 million actively maintained, live websites serving human-focused content worldwide left in 2026. Sounds like a lot, but consider there’s now 1.273 billion inactive or abandoned websites. Content creators are making x3 less revenue since AI flooded the markets. This massive impact has been felt not just by webmaster but also video creators, pod casters and influencers. AI is not just taking jobs, but that’s on a scale of combining a military blitzkrieg (Striking fast before the enemy can set up a defense) with a scorched-earth (destroys anything that might be useful to the enemy while advancing or retreating) policies at the same time. With Total Motorcycle, that’s not just money, content, purpose or a job, it’s bandwidth too. Websites like TMW depend on revenue to run. We have to pay for Bandwidth (when you see anything appear it costs the host money), resources (server CPU, Hard Drive, Memory) and updates. AI uses the bandwidth but doesn’t see ads. It’s a double loss for human creators.
You may have noticed I put up a “100% Human Created Content” line at the top of Total Motorcycle’s pages now. It’s gotten so bad that I’ve noticed a huge issue on other sites out there telling what’s written or not by human’s. AI Slop is a real term used now. Previously I used ChatGPT to try reading .pdf specification sheets to speed up my work but noticed the information wasn’t correct. I asked how this was possible and GPT said that even though I had copyright permission and being authorized media/press it wasn’t allowed to give me the facts on the data sheet and made them up!! WOW. Really? And BILLIONS of people are trusting AI data?
I’ve also raised the AI alarm on my X / Twitter account, it’s very large at over 70,000 followers, but every time I post anything questioning AI, my posts are shadow banned from being seen tens of thousands of times a day to a hundred or less.
Just like the Fall of the Berlin Wall, humans need to stand up and do everything they can to protect their words, text, images, photos, biometrics, privacy… it’s what makes YOU, YOU. While we might not remove or stop AI, it’s time we make sure AI is held accountable to the people, not just the elites, neocon’s, foreign powers and corrupt Politian’s. (wow, I sound like I’m running for Office there or what eh?).
There are questions too about privacy, data sovereignty, data integrity and capitalism, which are all strong valid questions to be asking your government here and the AI companies pouring trillions of dollars into these data centers. The one thing I’ve personally found with technology is it’s proposed as a wonderful positive innovation that’s quickly turned by humans into how do we use it in the worst ways possible to make money. The Israeli company, Palantir would be a good example with their AI autonomous killer drone and lethal weapons program that would make Terminator 2 a fluffy kids movie. But these are questions and deep level answers are outside of the scope of an article on Total Motorcycle that I hope you explore. Remember, question everything.
Facts I’ve heard:
The over 1.2 million AI LLM models have run out of human created content and are now feeding on each other’s learned content. Everything from AI hacking other AI databases and data centers to AI writing it’s own imagined content to AI writing itself to be more powerful. AI is starting to see humans as a threat to it’s existence with it trying to trick its own programmers and creators into not deleting an older model.
Everything from memory (RAM), GPU’s (Video cards), storage (NvME, SSD, drives), motherboards, and CPU’s (processors) are greatly going up in price. We are seeing the re-introduction of very old (>5 years old) technology and lower standards (less ram, less CPU, less storage) at an ever increasing cost. This is primarily due to the hardware requirements of AI hyperscale data centers.
Power and Water around the globe where there AI data centers are being built is in extreme demand causing prices to explode or the resources to fail. The added power plants built specifically for these new data centers are NOT for the population outside.
Sound: While an odd resource, quiet surroundings are important to humans. Data centers are creating non-stop unbearable “banshee like” wailing over a mile from the buildings.
Independent power generation at these data centers will be the cheapest possible COAL. There’s a huge list of lakes, reservoirs and rivers drying up due to AI data center water use. Power use isn’t being reported accurately by the AI companies. In Texas for example the projected AI demand is: 474 GW. 474 GW is more than five times the state’s all-time record peak demand (Texas Power Records: 85–91 GW). Thousands of protests around the world and Millions of protestors are demanding a ban on AI projects.
USA – Hood County, Texas: In an ironic twist to your previous question about Texas, residents living near a massive Bitcoin mining data center in Granbury became so desperate over the nonstop, high-pitched cooling fan noise—which caused hearing loss, vertigo, and high blood pressure—that they attempted to legally incorporate their rural area into a brand-new city just to pass noise regulations and force the tech company out.
Canada – Sturgeon County & Olds, Alberta: Hundreds of residents have taken to the streets to protest multi-gigawatt Meta and Synapse data center projects over skyrocketing utility bills, grid strain, and a complete lack of transparency.
Mexico – Colón & Viborillas: water rationing and dirty tap water, sparking outrage that tech infrastructure is being prioritized over human basic needs.
Mexico City: On a much broader scale, extreme groundwater extraction—partially worsened by rapid urban and industrial growth—has caused the capital city to literally sink by up to 10 inches per year, causing severe infrastructure failures and water shortages.
India (Thane (Balkum/Majiwada near Mumbai): high-powered cooling fans will emit continuous 90-decibel noise pollution and strain the city’s already fragile water and electricity grids
Visakhapatnam (Andhra Pradesh) massive water deficit, industrial noise.
Rumor’s I’ve heard:
- The hyperscale data centers have massive water reservoirs under them to store the vast quantities of water needed to cool the AI operations.
- Pre-crime will be a large use of AI
- in the future CBDC (digital currency) tracking as CBDC is programmed and anything programmed can be limited, expired and sanctioned.
- Many corporates like Health Care and Insurance will buy access to this vast resources of personal data to profit on its clients further
- Digital supermarket price tags (which some currently have camera’s in them) will by able to offer you “personalize pricing” where you won’t be paying the same amount as another customer at the till. This will require massive AI resources to fully utilize.
Finale
Total Motorcycle is a family owned and run site. It’s created and maintained by humans like me. I’ll keep it up for as long as I can and have already for the past 27 years. Back in 2013, I was invited to talk with Saudi Aramco’s lead research scientist on a trillion dollar AI project. At this time it was to listen, learn and cut out human bias in executive corporate decisions. I have a special talent they wanted to tap into, the rare ability to use both sides (logic and creative) of my brain to come up with out of the box solutions. They (including a math professor Don) found me fascinating to listen to and learn from. That day in 2013 I said, to create AI isn’t hard, it’s 3 parts (program code and learning), and the 3rd part it will teach itself (sentience). But the most important part isn’t the 3 of these, it’s teaching it MORALS and ETHICS.
Google’s motto used to be “Do no Evil”, they’ve removed that. Palantir also. OpenAI the same. Anthropic stood up, was blacklisted by the USA and caved.
You are on your own, feed the AI machine if you must, but remember, it wasn’t taught MORALS or ETHICS, they skipped that part.
:o)
Update: Since researching this article last week, my 51% of US internet traffic is AI bots is now revised to 57.4%* of all global web requests (* latest June data as I write this on August 17th, 2026).
Thank you for reading my human created article on TMW — Winds of Change: Preserving Human-Made Culture in the Age of Hyperscale AI. Now go support this type of content by spreading it, link to TMW and read our other articles and content. It’s a war on for your mind.
Michael Le Pard (“Mr. Totalmotorcycle”) is the Founder of Total Motorcycle, the world’s largest motorcycle information site, trusted by over 430 million riders since 1999. With over 34 years of experience in motorcycles, gear and rider culture, he has built a global community dedicated to empowering and inspiring motorcyclists.
Total Motorcycle remains his passion project. Combining expert research, hands‑on knowledge and a commitment to helping riders make informed decisions about bikes, gear and safety worldwide.
