At the end of every year, I listen to and read about others predictions for the coming year - technology, games, etc. This time, I thought I'd throw my hat into the ring. It's likely I'll get most, if not all, of these predictions very wrong.

Tech

A new job title in tech

With LLMs becoming more useful with each model release, the barriers between product management, design, and engineering are becoming much blurrier. In 2026, there will be a job title, new or repurposed, that has a list of expectations that's a chimera assembled from product, design, and engineering.

This prediction will be considered correct if at least one FAANG or LLM model company posts job openings with a new job title we haven't seen before or was previously under-used (design engineer, product builder, etc.). The job description must specify how LLMs will be expected to aid in the job and how the job responsibilities differs from what may be expected of roles with similar names.

Figma share price continues to go down

As a designer, I use Figma daily. LLMs have steadily cut that usage. Why? I can hop on ChatGPT, Claude, or Gemini, and pump out real, interactive, prototypes that I can quickly iterate on with a prompt. No drawing boxes, no dealing with finicky interfaces - just real, interactive, results. It's hard to compete with that.

Figma is trying to battle this move with Figma Make, but I don't believe it'll be as good as many of the other core model offerings. This shift will directly result in the continued devaluation of Figma.

This will be correct if the Figma share price is equal to or lower than what it currently is ($37.63 as of this writing) at the end of 2026.

LLM model performance flattens out

I'll admit - I have a very high-level understanding of how LLMs are made. So I'm pretty sure I'll be wrong here. I'm making the prediction anyways.

Models will reach a plateau of real-life useful performance. New models will be released, people will use them, and they'll mostly do the same things, to the same level of correctness, as they did before.

Some of this happened in 2025. A thousand thousand voices seemed to speak in unison "meh" when ChatGPT 5 was released. However, models like Claude Opus 4.5 and Gemini 3 seem to do much better than previous versions. I personally find Opus 4.5 to be much better than Sonnet - it's the model I use on a daily basis.

Vibes and my own personal experience will determine if I get this predication correct. Less quantifiable than the other predictions, but more fun for future me to rationalize.

Gaming

The Steam Machine is super expensive and it sells anyways

GPU and memory prices are going through the roof and tech components as a whole becoming more scarce. One of the most anticipated console releases of 2026 (if you can call it a console), the Steam Machine, will be extremely expensive. And a lot people will buy it anyways.

This prediction will be considered correct if the lowest-spec model for the Steam Machine is priced at $999.99 and 500k or more are sold, in total, within 2026.

Capcom resurrects the Breath of Fire series with a new, standalone, single-player, RPG title

With Clair Obscur: Expedition 33 winning game of the year (amongst other awards), the time for investment in otherwise dormant turn-based RPG franchises by larger companies is now. Also, I really love the Breath of Fire series and want more games to be made in it.

Success means an announcement of a new Breath of Fire game in development for major consoles and/or PC. If these even happens, I highly doubt we'll see gameplay or a release window. Remakes will make this prediction correct, but only if they are a complete remake of a previous game - not a remaster. More FF7: Remake rather than FF7 released on PC.

Markets

Prediction Markets Rise in Popularity

I'm convinced the term prediction markets is a clever guise for gambling. Despite the damage they'll cause, the US government will take no action against these platforms and, if anything, will continue to lessen any restrictions they may have in operating.

If prediction markets continue to gain in popularity, defined by total volume of all prediction markets, in 2026, this will be considered correct. I really hope I'm wrong here.

Electricity prices spike - and then go down

This is a prediction related to the LLM model performance. In short, if model performance flattens, investment will flatten. If investment flattens, the need for large-scale data centers will go down. Data centers need a lot of electricity, some of which has been pre-purchased by these tech companies. If they don't need that electricity, they'll likely sell it back to the grid to recoup losses. The more electricity available in the grid, and lower demand given less data center usage, the lower the price. At least I think that's how it works.

I'm not super confident about this prediction for two reasons: Even if OpenAI and other companies were to stop their data center buildouts completely in the middle of 2026, I'm not so sure the electricity they've already purchased is enough to maintain whatever they would have built. There's also the problem of even if LLM model performance flattens, some of these companies may continue to pursue scale in an attempt to create AGI, despite signals that LLMs may not be the right path towards that goal.

If aggregate electricity prices in the US are lower in December of 2026 than they were in December of 2025, this prediction will be considered correct.