🌊 The Silicon Rust Belt

Did someone forward you this? Subscribe here free!

By Barratt Dewey

“Learn to code,” they said.  

“Computer science skills would lead to a stable, lucrative, and future-proof career,” they said.

Not anymore. 

In the past two years, more than a quarter of all computer programming jobs have vanished, the worst downturn in industry history. Fewer programmers are now employed in the US than at any point since 1980, despite the US’ total workforce growing 75%. The decline started around 2022 – the year OpenAI released ChatGPT. 

Could computer programmers be the first casualty of widespread AI adoption? In this deep-dive, we examine the trends and data to see if we’ve passed peak programming and if the tech-job boom is vanishing before our eyes. 

Just three years ago, it was a programmers’ paradise. Covid-induced remote work sent tech stocks soaring in 2021 and 2022, fueling the most aggressive tech hiring spree in history. 

Between 2019 and 2022 alone, Amazon’s staff grew by 106%, Meta’s by 103%, and Google’s by 64%. While employee counts and salaries exploded for the cash-flush tech companies, the workload did not. Throughout the pandemic, videos posted by tech workers went viral, describing little work and high six-figure salaries. In 2021, entry-level programmers at Meta were paid between $189,000 and $291,000. Google paid a similar amount. 

As one ex-Big Tech employee told Roca, “Most people weren’t doing shit.”

Many accused tech companies of hiring as a “vanity metric,” so it looked like they were growing, or to prevent qualified talent from working for rivals. Many employees agreed: “They were just kind of, like, hoarding us like Pokémon cards,” said another former Meta employee in a viral TikTok. 

That era is now over.

Since 2023, programmers have been laid off at lightning pace. In 2024, there were 35% fewer software developer job listings on Indeed than five years ago, and the tech industry laid off over 152,472 employees. 

Tech layoffs have continued to accelerate this year. One cause is the declining demand for tech services post-pandemic, but another is technology itself. 

AI tools are able to do significant chunks of programming work. Since ChatGPT was released in late 2022, hiring for AI-specific roles has risen 68%, while tech industry-wide job postings have fallen 27%. AI engineers now account for nearly a quarter of all tech hiring. 

These trends are mirrored in salaries: AI engineers are being offered salaries up to four times higher than regular engineers, and in 2024, non-AI engineer wages increased by an average of just 0.95% – less than the 3.8% average across the US workforce.

So is programming no longer a sure path to a high salary? 

The data suggests that the glory days for America’s coders have ended. Hiring for software engineers has dropped 70% from its 2019 peak, and those once-lucrative salaries have flattened. By the end of 2023, tech recruiters were often offering compensation 15-25% lower than the “glory days” the year before. 

For computer programmers, those who primarily write code, this decline has been more pronounced. In 2018, there were about 250,300 programmers employed in the US. By 2023, the Bureau of Labor Statistics (BLS) estimated that number had declined roughly 45%. It projects a further 10% decline in programmer employment in the coming decade. 

Many of the reasons cited for the decline have nothing to do with AI. Some argue that the tech workforce was hyperinflated during the hiring boom in 2021-2022 and thus had a longer way to fall as it corrected. Others point to outsourcing entry-level coding tasks to foreign workers as the primary reason for the drop. However, this trend appears to be amplified by AI replacing those outsourced workers.

If computer programmers are concerned about AI replacing them, they aren’t showing it in their work. Ironically, computer programmers are incidentally training the AI software by becoming increasingly reliant on it. Roughly 60% of developers surveyed in May 2024 by StackOverflow, a software developer community, said they had used AI coding tools in their jobs. Another survey on GitHub last year found that 97% of software engineers are using AI coding tools in some capacity.

Many programmers and tech industry leaders argue the concerns of programmer extinction are overstated and that employees are using AI tools as a supplement rather than a replacement. Google’s CEO Sundar Pichai recently said that more than a quarter of the company’s new code is now generated by AI before being reviewed and approved by humans. IBM’s CEO, meanwhile, estimated that AI could feasibly write 20–30% of code in the near term, handling simpler cases, but said he believes more complex tasks will remain entirely human-driven.

The AI industry disagrees. OpenAI CEO Sam Altman stated that “the world’s best programmer will be an AI by the end of 2025.” AI programmers, he said, could work 24/7 at a fraction of a human’s cost​. Anthropic’s CEO, meanwhile, recently predicted that “90% of code may be written by AI in the next 3-6 months.” While these statements are dramatic – and both CEOs have a financial incentive to make them – they could be right, though perhaps not at the scale and imminence they claim.

Rather than complete automation, many in the tech industry say a more likely outcome – at least in the short term – is that the nature of coding jobs will change to adapt to AI, which has dramatically increased the productivity of programmers. One recent study found that using an AI coding assistant made developers 20% more productive. Another study found that software developers with access to GitHub’s Copilot chatbot finished a coding task 56% faster than those who did it alone. One programmer who worked on Roca’s app estimated that the app could have been built 50% faster had versions of ChatGPT been available. 

While advances in technology have typically hit low-skilled workers the hardest – automation for factory workers, for example – AI’s impact could be greatest among white-collar workers, including software developers, lawyers, and accountants. As AI improves to perform more sophisticated work, those more educated workers could find themselves on the chopping block.

“Without getting hysterical,” Mark Muro, an AI and workforce analyst at the Brookings Institution, said, “the unemployment jump for programming really does look at least partly like an early, visible labor market effect of AI.” In an Anthropic-led study of 700 occupations that overlap most with AI’s capabilities, coding roles dominated the top 10 by a large margin. 

However uncertain the future may be for programmers, the broader adoption of AI could also create entirely new jobs in the tech field. In recent years, the number of job openings for AI prompt engineers and curators has surged, leading analysts to project that most software workers will soon be supervising and refining AI-generated code rather than writing it themselves. No matter what ends up happening, one thing is clear: AI’s impact on the workforce is only going to grow. For America’s tech workers, the coming years may be defined by one motivation: Adapt or die.

Editor’s Note

Thank you for reading. We’d love to hear your reactions to the shift in computer programming jobs. Will AI help or hinder the sector? For the coders in the audience, has your job been impacted in any way so far? Send in your replies here! 

In case you missed our most recent stories, find them here: 

• The legality of a third Trump term

• The Florida Blue Key’s takeover of DC politics

• Battling for chimpanzees in Sierra Leone

• The truth about the EPA’s $20B “slush fund”

• Are Republicans in trouble?

Lots of replies to yesterday’s story on Trump potentially running for a third term. Here are a few:

Caroline wrote in:

Wanda disagrees:

And Heidi from Maryland wrote:

That’s all for today. See you tomorrow.

–Max and Max