We are coming to the end of 2025.
This year saw a massive move forward. As they say, you ain't seen nothing yet.
We are likely to see a larger advance in 2026. There is an acceleration that will take place. For the moment, we are seeing most asleep to this fact. My guess is that 2026 will be the year that many people start to catch on.
AI is taking over. We are seeing math problems being solved. Coding is stepping forward. Models are getting better in every area.
It is interesting to watch how the defenders of the "non-AI" status quo keep moving the goalposts.
There is a multi-trillion dollar question on the table.
Can we scale compute into new discoveries? This is what many in the area of AI are focusing upon. It is one thing to run a chatbot that serves as advanced search. Learning techniques that humans use is also a major advancement yet still mimicking humans.
Will we see AI develop something completely new? Can the transformation of a certain amount of compute lead to discoveries that never were seen before? Does AI achieve the status of advanced researcher or scientist.
One area to watch is the field of math. This is one area where people are starting to take notice. The math community, in the last few months, is starting to pay attention. Many are embracing the fact that, in the not-too-distant future, AI will be able to solve some of the hardest math problems that have plagued mathematicians for centuries.
This is only feeding into the belief that acceleration is upon us.
One of the key realizations for this past year was the fact that a lot more intelligence can be created after training. Quantization is now yielding great results. This could provide 100x parameter breakthrough over the next year.
We are seeing the chips getting more powerful, with more compute being added. Budgets for Big Tech investment is only getting larger. Again, this is not something that is going to slow.
China is locked out of the chip game. For this reason, it is diving deep into quantization of its models. Deepseek stunned the world a couple years ago. The advancement of models using smaller parameters is happening.
This will have a global impact since most of what China is doing is open source. That means it will flow to different parts of the world, especially to the United States. It will, however, negate the chip advantage the US is intent on maintaining. China might develop their chips to handle the quantized models, providing more efficiency against the US counterparts.
This model is likely dead.
Instead, we are looking at AI replacing major sections of companies. That means no transformation but, rather, a new organizational chart. Those companies that make this switch will be able to keep ahead of their competitors.
Everything will incorporate AI. If the predictions of some hold true, that we will see a 100x step in capabilities, then the world will be a different place. A couple years of this type of advancement becomes mind blowing in a short period of time.
Entire companies will be built on this concept. Just like Google was an Internet company, not existing before that innovation, AI companies will spring up and take over industries. Some established companies will be disrupted. However, the major thrust forward could come from entirely new industries that AI focuses upon.
The bottom line is that everything is moving towards digital. In just a few years, there will not be a choice. It is akin to not having an online presence today. What business could survive like that? Not too many.
Platforms will be the new norm. This can come in a variety of ways, with delivery systems expanding. Those which focus upon this can realize the gains where others miss. The acceleration is taking place, likely leading to the world waking up.
Do not state that you weren't warned.