Predictions for 2025

TLDR

1. Silicon Crisis II
2. Video Generation Golden Era
3. StarLink Competitor
4. Strong Australia / US Alliance

Why?

My original source of inspiration for this post is LaurieWired’s youtube video where she explains the importance of making predictions. I also think it’s important (and fun!) to just take a step back to make some predictions and see how they play out.

One important factor to consider is that I am currently a CS dev, which means my predictions are mainly going to be about technology and such.

Like LaurieWired, I’ll be sharing what I think is the probability of each event occurring, as well as verifiable success/fail statements and explanations.

Without further ado, here are my predictions.

Prediction 1: Silicon Crisis (50% chance)

NVIDIA stock price increases by 20% ($120) due to unprecedented demand of Silicon Chips

This prediction is based off of the 2020-2023 chip shortage which showed just how monopolistic TSMC is. I think Trump’s strong stance on China, combined with the volatile nature of Tech/AI, will increase the chances of a major ‘trigger’ that causes TSMC to either increase manufacturing prices or greatly decrease production.

However, there are some notable anti-factors:

1. Alternative chip manufacturing plants (70% chance)

The 2022 CHIPS and Science Act has shown that the US government is keen to move chip manufacturing in-house, and while I think the act was not that successful, there is a good chance that this trend will only continue.

1. Trade agreements with China are met (50% chance)

I think it is likely that the US government will likely directly involve China in resolving the Taiwan issue. I don’t know exactly what it could entail, but I believe a diplomatic solution is more than viable.

1. AI slow down (10% chance)

If the machine learning space slows down, then it makes sense that global demand for silicon chips would decrease. This would happen if there are no good advancements made in the next year, which i doubt due to shear amount of potential and investment in the area.

Overall I think a 50% chance of a ‘silicon crisis’ is a good probability given that it has happened before and that I believe the underlying issues have not been fully resolved.

Prediction 2: Video Generation Golden Era (20%)

A popular movie (>$100 budget) comes out with AI generated scenes / actors as a selling point

I believe that AI video generation is currently at the “GPT-3” stage, where the majority of advancements are made at the high-end business level. However, as interest increases, there should be innovations that are made in the research space that help increase performance and make it feasible to run on consumer hardware. At that stage, performance should reach a point where it is feasible to use generated videos in a commercial setting.

Currently, the most feasible solutions are Mochi 1 / Sora / MiniMax / Hunyuan, which perform well, but I think the biggest limiting factor are copyright ‘handcuffs’, which prevent any real investment into the area. I think I Chinese competitor like DeepSeek would have the most success in this regard, but it also heavily depends on the abilities of the generated video and if it’s marketable / better than current classic film making techniques.

However, if this challenge is overcome, then there should be more interest in the field, resulting in a ‘cascade’ effect where more investment and development occurs.

I chose a 20% chance of this event occurring because it really depends on how copyright issues are dealt with and if a company is confident with the risks.

Prediction 3: StarLink Competitor (5% chance)

A company launches more than 50 satellites for communication purposes

I think StarLink has shown itself to be extremely useful when other providers go down during natural disastrous or government intervention. However, while I think a competitor to StarLink would be extremely valuable, it is also unfortunately unlikely as it requires a lot of capital investment, which is why it might only be possible as a government contract.

The reason why I chose 50 satellites is because current competitors to StarLink only have 3-5 satellites in geostationary orbit. While this does provide global coverage, they are placed far enough that they can only be used for extremely low bandwidth and so only can be used for emergency purposes. Alternatively, something like Iridium or GPS uses closer to 50 satellites for a faster yet still global coverage, so that is my threshold for ‘success’.

Of course, I only chose a 5% chance because launching satellites is extremely expensive, and so I don’t think this prediction is likely to succeed.

Prediction 4: Strong Australia / US Alliance

90% chance that the United States does not drop AUSFTA

Currently, the United States and Australia have a free trade agreement AUSFTA, where >95% of trade is traffic free. I believe that this should continue in the near future (despite Trump’s tariff hunger) and would be very surprised if this changes during his presidency. The only thing that might cause my prediction to fail is if there is strong anti-american sentiment due to radical progressions in geopolitics, which I think is very unlikely.

Another factor is that Australia exports a lot of simple commodities such as natural ores / minerals and textiles. One interesting fact is that we export more Education than Gold, which I think might change in the near future (~30%).

It is also important for us to be on good terms with the United States because we are a relatively small country, and our currency is very much tied to the USD. I think this will prevent us from becoming more independent like Switzerland, but in the long term I would not be surprised if we develop strong global financial institutions.

Ultimately, I think 90% is a very reasonable probability of Australia / US furthering in trade & technology cohesion mainly due to our natural resources.

Wild Cards

These predictions are what I would consider ‘within the realm of possibility’, but are still noteworthy. I placed them as wildcards because I don’t think I have the experience to determine wether or not these events are likely to occur, and certainly not to fully explain my reasoning behind them. Instead, they are more for bonus points if they happen at all.

Netflix incorporates AI upscaling features

Quantum computers are ‘miniaturized’ (still very expensive)

Bitcoin reaches at least 120k

OpenAI progression slows / encounters legal troubles

Hard British recession

Artemis II succeeds primary mission objectives, but is delayed

Germany leaves the EU

Meta Notes on Predictions

I took major inspiration from hypertextgarden’s blog post on making predictions for 2025. While it may seem that her predictions are the ‘meat’ of the blog post, looking back, making predictions is difficult enough as it is, which is why I want to spend some time on several epiphany moments I had.

My biggest discovery is that making predictions is really, really, hard. While I share a lot of conclusions with LaurieWired (my inspiration blogs author), I think she undersold the effects of hindsight bias when directly trying to predict the future; not just for personal validation on historical events

For example, lets take the ‘prediction’: “The South Korean President, Yoon Suk Yeol, is extremely unpopular, so he might be impeached, but he might declare martial law, which results in the devalue of the nation’s currency”.

Here, I think hindsight bias works against us as we could take any other random topic and make a similar drawn out conclusions, like “the ukraine-russia conflict may end as it seems to be slowing down, which would allow for more ukrainian wheat manufacturing, which leads to cheaper bread, which might cause countries to implement tariffs on foreign food imports to protect local industries, which ….” on and on and on.

In my personal opinion, the hardest part about making predictions is trying to avoid these long chains. While each step might seem logical, it’s a fallacy that hides the true probability of each stage.

Another thing I found was a bias towards negative events. At first I thought that it might be because in general people value preventing bad things from happening, but I also think it’s because they tend to be extremely unpredictable. Lets take for example World War 3 vs A Major Quantum Breakthrough. Even if we assume that both events are ‘equally likely’, there are a thousand ways WW3 can start, but there only needs to be ‘one’ quantum breakthrough for the other event to succeed. All of a sudden you have rely on your personality and what you value more in order to determine which event is more likely.

Conclusion

I wholeheartedly recommend taking the time to write down your own predictions - not just for personal reflection, but to train your creativity and thoughtfulness.

Even if your predictions are big or small, unlikely or certain, I think they show a lot about what you value and how you view the world, and so it’s a great time machine into your past.

I am fairly confident with my predictions, but I am more worried about what I missed rather than kept. I am sure that there are painfully obvious factors that I missed, but there was simply no way of knowing if it was important or not.