Ferminius MMXXVI Book a diagnostic →
Ferminius/ Journal/ The future of labor demand
2026.09.29 · Note 06

The Careers I Bet On in 2020, Checked Against 2026

In February 2020 I published a short piece on this site about the professions that would pay off in the new decade. Solar and wind technicians, big data people, SEO specialists, programmers, VR developers, business intelligence analysts. A month later the whole world went home to work from the kitchen table, and three years after that everybody was talking to a chatbot.

So I went back and read my own article, slowly, with a coffee and a little bit of shame. Some predictions held up and I missed the biggest story of these six years, and a couple were already wrong the day I wrote them.

The employment numbers below come from the Bureau of Labor Statistics Occupational Outlook Handbook, updated on August 27, 2026: projections for 2025 to 2035 and median pay for May 2025. For reference, the median wage for all U.S. workers is $50,980 and all jobs together are expected to grow 3%.

My 2020 text said solar installers and wind turbine technicians would be the fastest growing jobs in the country. They are still near the top. Solar photovoltaic installers are projected to grow 37%, with a median pay of $53,140. Wind turbine technicians sit at 30% and $64,120.

Back then I left out the size of these occupations. About 31,100 people work as solar installers in the whole country, and the BLS expects around 11,300 new positions in ten years. It's still a good trade with real demand, and you can enter with a high school diploma and training on the job, but you will be competing for about 4,200 openings a year.

Big data and business intelligence I treated as two separate careers. Today most of that work lives under one title, data scientist, and the BLS tracks it directly. Projected growth is 35%, third on the whole list, with a median pay of $120,230. My 2020 piece also said these people worked mostly in HR, marketing and sales. I would delete that sentence today, because I had nothing behind it.

A confession while I'm here. In the original I gave a "96% upward trend" for SEO specialists. The BLS has never tracked SEO as its own occupation, and today I can't find where that number came from. I think I copied it from the wrong line of my notes, and it stayed on this site for six years.

Hands in handcuffs resting on papers on a desk, with a night city skyline and bokeh lights blended over the photo
A confession · the 96% SEO figure had no source

Programmers went the other way

In 2020 I called the programmer a kind of rock star and wrote $100,000 a year, citing PayScale. The BLS median for computer programmers in May 2025 was $100,390, so the pay held. The outlook is worse: employment is projected to shrink 7% by 2035, and the BLS expects companies to use AI to automate the repetitive programming tasks.

Two software developers look at a computer monitor in an office, one resting his chin on his fist
Software developers · +10% to 2035, median $135,980 (BLS)

Software developers are growing 10%, with a median of $135,980, and the BLS expects about 106,100 openings a year across developers, QA analysts and testers combined. They start from the same technical base as programmers and then decide what gets built and how the pieces fit together. If you learned one language in a bootcamp in 2020 and stopped there, these numbers should worry you a little.

What SEO looks like for me now

Growth Marketing Studios' free AEO Grader page, headlined "Most Florida companies are invisible to AI search right now", with a form to test a brand in ChatGPT, Perplexity, Gemini and Claude
Answer engine optimization · GMS AEO Grader, four AI engines in one scan

This one is personal, so read it as one operator's experience. I have been doing search marketing in Miami for more than ten years. In 2020 the work was keywords, links and technical fixes so Google could read the page. I still do that part every week.

People now ask a lot of their questions somewhere else. They go to ChatGPT, Gemini, Perplexity or Claude, and the assistant answers directly, many times without sending anyone to a website. So a good part of my week is answer engine optimization: making sure that when an assistant explains a topic in my clients' field, it finds pages from them that are easy to quote.

A Public Adjusters case

In 2023 we took over the marketing for Alconero Public Adjusters, a public adjusting company in Florida. At the start, a team of six people handled the whole account. From day one we began training agents, first on OpenAI and later on Claude Code. Today the account runs with two human supervisors and several "adversary" agents that review everything we deliver to the client: content, social media and strategic planning. We build these agents for each specific client, inside the automations we deliver to our users. For the next few years AI will replace only part of the staff, and right now that part is the junior layer of the company. That raises the barrier to entry in fields like marketing and technology.

A plastic surgery clinic

A few years ago we managed the account of a plastic surgery clinic. After a tragic accident, a patient died. A family member left a comment on a review, and an AI agent answered it as if it were any other review. The reply was completely out of place, and the incident turned very negative because an agent without the right training had stepped in. The situation was eventually resolved, and it changed the way we work. Since then, every agent we train, especially the ones in direct contact with the public, goes through extreme scenarios, pressure and stress before it touches a client account.

Anyone starting in SEO now has to learn to write for Google and for the assistants at the same time, and in my experience most courses still teach only the first part.

VR, where I got excited too early

I wrote that every day more companies were committing to virtual reality. BLS doesn't track VR developers as a category, so any number I gave you here would be made up. My read, from the agency side: VR found real niches in training, architecture and some medical work, and that is about as far as it went. If I were choosing a career in 2026, headsets would be a side skill for me at most.

A businesswoman in a brass helmet with a light bulb and dark goggles holds a wrench next to a cardboard robot
Early enthusiasm · VR stayed in training, architecture and medicine

The part I missed

My 2020 article ended with Deep Blue and Kasparov, and I even got that story wrong. Kasparov won the first match, in Philadelphia in February 1996, 4 to 2. Deep Blue won the rematch in New York in May 1997, 3.5 to 2.5, according to IBM's own history of the project. My 2020 version mixed the two matches into one.

The next step arrived much faster than I expected. OpenAI released ChatGPT on November 30, 2022, and in a couple of years language models were writing code and answering the questions people used to type into Google. Data scientists sit third on the current BLS fastest-growing list, and information security analysts (21%, median $129,180) are on it too. Programmers are the one group from my old list that the BLS now expects to shrink, and its own explanation points to AI.

If someone asked me today

Pick a field where the work touches something physical or human, because that part is harder to automate. A solar installer has to be physically on the roof, and a nurse practitioner has to be in the room with the patient, which may be part of why nurse practitioners top the BLS list at 41%. Then learn to use AI inside that field.

The BLS handbook is free. Look up your own field before you listen to me or anyone else.

About the operator.

This page was written by Fermin Fleites — known professionally as Ferminius — AI marketing director and SEO consultant in Miami, founder of Growth Marketing Studios and Digital Media 305. Sixteen years of SEO experience. Máster de Inteligencia Artificial from BIG school, Barcelona, 2026 (240 hours), and Experto en Inteligencia Artificial certified by the Universidad Internacional Isabel I de Castilla, 2026 (150 hours, 6 ECTS). Read the full About page →

Reference · BLS figures cited in this note
OccupationGrowth 2025–35Median pay, May 2025
Nurse practitioners+41%$132,300
Solar photovoltaic installers+37%$53,140
Data scientists+35%$120,230
Wind turbine technicians+30%$64,120
Information security analysts+21%$129,180
Software developers+10%$135,980
Computer programmers−7%$100,390
All occupations+3%$50,980

Source: U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, updated August 27, 2026.

Rethinking a team around AI agents?

30 minutes. Bring the org chart.