“Happy 2025. AI took my freelance writing gig,” declared YouTuber Alex Wei in a video that blew up on New Year’s Day. He opened up about how a client dropped him, opting instead for an AI chatbot to churn out blog content.

His raw question hung in the air: “How can I even compete with that?”

For Alex, and countless others in similar shoes, figuring out how to stay relevant is a real head-scratcher—even for those who are tech-savvy enough to use AI themselves to avoid being replaced by it. And even for those managing to ride the AI wave to keep their jobs, it’s becoming increasingly costly and tougher to hang on, especially in developing countries.

The latest “pro” subscription to OpenAI now hits $200 every month. RunwayML, a top video generator, asks for $95 monthly for its premium perks, while the best Midjourney (AI image generator) plan sets you back $120 a month. It’s a stark jump from just a year prior, when ChatGPT Plus’s top tier was around $20 and Runway charged $15 to access its Gen-3 video magic.

While $200 might seem like a workable business expense in the U.S., consider that it represents about two months of average minimum wage earnings in Venezuela. It’s also equivalent to two weeks of pay in Mexico or Chile, and an entire month’s minimum income in Suriname.

Even in booming economies like China, where minimum wages vary by region from $275 to $370 monthly, these subscription fees can eat up a significant chunk of a worker’s pay – especially for freelancers.

The AI haves and have-nots

These surprisingly high prices are drawing a line between those who can tap into the power of AI and those who are left watching from the sidelines. Daniel Vasilevski, who heads up Bright Force Electrical in Australia and spends $120 a month on Midjourney for his business, gets the bigger picture.

“The big thing I’m seeing is that AI is likely to widen the divide between companies that can afford it and those that can’t,” Vasilevski shared with Decrypt. “Businesses investing in advanced AI will pull ahead by automating tasks, better assisting customers, and making smarter decisions. Meanwhile, smaller players or individuals unable to afford it will find it much tougher to stay competitive.”

Vasilevski emphasized, “If access boils down to budget, then those with deep pockets will hoard all the advantages, leaving everyone else behind.”

Jeff Le, formerly a deputy cabinet secretary for California who managed emerging tech for Governor Jerry Brown, sees echoes of the past in today’s situation, but remains cautiously optimistic about what’s ahead.

“These AI tools have the potential to revolutionize how we work and open doors for greater innovation. However, it still feels early days and these powerful tools are mostly in the hands of a select few,” Le told Decrypt.

The concept of new tech concentrating wealth and influence isn’t new. The Gini index, which measures inequality between rich and poor, has generally increased over time. Even with the internet driving global GDP growth, the Gini index still climbed, indicating that the gap in opportunities and wealth between wealthier and poorer nations actually widened.

Basically, technology made countries richer overall, but it didn’t necessarily lift up the poorest populations. While globalization and new technologies boosted GDP collectively, a larger share of the profits went to a smaller group of people—amplifying the divide between the rich and poor.

Can regulations brook the divide?

This situation mirrors what happened after the Telecommunications Act of 1996 in the U.S., where market forces favored urban and higher-income areas over rural and low-income communities. By 1999, a mere 9% of U.S. classrooms had internet access—typically in wealthier school districts—leading civil rights icon Jesse Jackson to decry what he termed “technological segregation.”

U.S. lawmakers are paying attention this time around. A recently formed bipartisan House AI Task Force is looking into how to prevent AI from widening societal gaps, much like legislators tackled internet access issues in the ’90s. But just like in the early internet era, when an AOL subscription felt expensive, today’s AI tools come with premium price tags that could become increasingly out of reach as AI becomes more essential.

This could deepen the innovation gap. For example, AI-powered healthcare diagnostics are common in the U.S., but remain scarce in less wealthy areas due to high costs and limited data. Furthermore, regulations—like the EU’s AI Act—disproportionately affect smaller players, potentially holding back local innovation.

Of course, the situation might resolve itself over time. Academic researchers generally agree that while investing in AI might be a heavier lift for poorer countries initially, it’s ultimately beneficial in the long run.

“While catching up technologically is possible, it requires careful planning, investments in people, and smart policies,” according to a recent Nature study. “The lack of basic digital infrastructure, skilled workers, and research capabilities often blocks direct AI progress for low-income countries (LICs).”

However, “evidence suggests that technologies like mobile e-commerce and e-banking have taken off faster in low- and middle-income countries (LMICs) compared to high-income countries. This supports the idea that some LICs could potentially leapfrog in AI adoption if the conditions are right.”

Regulators may not have the last word

Without specific actions, like offering subsidized access to open-source models or hybrid cloud solutions, AI might just become another area of global inequality, echoing the exclusionary patterns of the early internet.

And some argue that this is a larger, systemic problem that rules alone can’t fix—that the market itself will find a balance.

Increased competition could indeed bring prices down. And open-source options, such as China’s DeepSeek R1, which outperformed even OpenAI, could level the playing field. Beyond its open-source nature, DeepSeek offers a language model for power users at a mere $0.07 per million tokens—a tiny fraction of GPT-4’s $2.50 price. The company served a clear warning to industry giants, suggesting that high prices are more about market dominance than actual costs for computing or R&D.

As a result, OpenAI made its powerful reasoning model available on the cheaper “Plus” tier, Perplexity adopted a local version of R1 for Western users and launched a deep research model, and reports indicate Anthropic is also developing a reasoning model to stay in the game.

“Market forces will tackle AI accessibility more effectively than government regulations,” Karan Sirdesai, CEO and co-founder of AI infrastructure firm Mira Network, explained to Decrypt. “More companies are creating open-source alternatives to premium AI tools, which builds competition that benefits smaller and medium businesses. This natural move towards accessible solutions mirrors how other technologies have become widespread through market dynamics, rather than just regulation.”

Even OpenAI CEO Sam Altman is exploring out-of-the-box ideas to promote AI access for those often overlooked:

“It really does seem like the balance between capital and labor could easily get thrown off, and this might need us to step in early,” he wrote on his official blog. “We’re open to even unusual ideas, like giving everyone on Earth a ‘compute budget’ to make it possible for them to use a lot of AI.”

This is still not a perfect solution, as it might just make users even more reliant on OpenAI’s tools, further solidifying the company’s dominant position. Whether open-source alternatives, regulatory actions, or simple market competition can level the playing field is still uncertain—but right now, the AI revolution is far from being evenly spread.

“Ultimately, regulation has to balance risk control with encouraging innovation, ensuring AI doesn’t become a tool only for the wealthy and powerful,” Atul Arya, CEO and founder of AI software company Blackstraw.ai, told Decrypt.

“We have to prioritize fair access to the infrastructure, expertise, and funding needed to build AI solutions,” he added. “Open innovation spaces, partnerships between public and private sectors, and initiatives to make it easier to start with custom AI development will be key to making sure AI’s benefits are shared broadly.”