DeepSeek: Open Source, Closed Wallets

The LLM landscape is evolving fast. Open-source models like DeepSeek are making powerful tools more accessible, while companies like OpenAI are introducing ultra-premium services like the $200-per-month ChatGPT Enterprise plan.

This combination—the availability of free, open-source models and high-cost premium services is forcing us to rethink how we value AI. The question isn’t just about price tags, though. It’s about what makes AI work in practice and where the long-term opportunities could lie.

Here are five theses on how the AI landscape might be transforming:

1. The Fundamentals Still Matter

At its heart, AI runs on three essential resources: computing power, data, and accessible systems for innovation. These fundamentals aren’t going away.

Yes, training and inference have become faster and cheaper. But the drive to create better models hasn’t slowed. Open-source AI tools are important breakthroughs, but they can’t escape these realities. The ability to access a model is one thing. The ability to apply it effectively is something else entirely.

2. Infrastructure Unlocks Potential

Models like DeepSeek can’t create impact on their own. They need ecosystems and infrastructure that make them scalable, reliable, and intuitive.

This goes beyond cloud giants like AWS and OpenAI. Companies like Apple and Microsoft are also laying the groundwork for tomorrow’s use cases and that is their moat. Their plans for device-level integration and seamless user experiences could be critical to making AI a practical every day tool.

DeepSeek, for all its promise, is missing some of these practical layers. Features like custom GPTs or polished UI tools are what make proprietary models so compelling to non-technical users. Without them, even the most advanced open-source model struggles to deliver value where it matters most: in day-to-day workflows.

That said, the pace of open-source iteration is remarkable. Communities of developers are working rapidly to fill these gaps, and it wouldn’t be surprising if DeepSeek—or something like it—soon offers the same ease of use as premium tools. For now, though, proprietary platforms maintain an edge when it comes to convenience and polish.

3. Applications Are Where the Real Work Happens

For most use-cases, the true power of LLMs doesn’t come from its raw intelligence. It’s found in how that intelligence is deployed.

This is why the application layer is so important. Many businesses don’t need the newest, flashiest models. Instead, they need tools that solve problems efficiently. Often, an older model combined with a well-designed interface is enough to deliver exceptional results.

This makes it an exciting time to work on developing use cases and applications that build on top of these foundation models. By focusing on specific industries and the problems you understand well, there’s a huge opportunity to create solutions that deliver real value.

4. Open Source Fuels Progress

Open-source AI is doing more than democratizing access. It’s driving innovation at breakneck speed.

By opening up these tools, developers can tinker, refine, and iterate faster than any single company could manage. The benefits ripple outward:

• Customers gain cheaper, better products.

• Developers enjoy the freedom to experiment and customize.

• The entire ecosystem moves forward, faster and more competitively.

But with this flexibility comes responsibility. Removing built-in constraints—like DeepSeek’s censorship on politically sensitive topics—allows for customization but also raises ethical questions. The challenge is ensuring that open-source freedom isn’t misused.

In the end, though, open-source AI’s collaborative nature makes it a powerful force. It challenges traditional providers, pushes down costs, and raises the bar for everyone.

5. Commoditization Reshapes the Market

The rise of open-source tools like DeepSeek is reshaping how AI is priced.

For many use cases, AI models are already “good enough.” This reduces demand for ultra-premium subscriptions, like OpenAI’s $200-per-month enterprise plan. Users will pause before paying that price, especially when free or low-cost models can do the job.

However, mid-tier offerings like ChatGPT’s $20/month Plus plan may feel less of a squeeze, as they strike a balance between cost and value.

For users, this is great news: AI tools are getting cheaper and better. But for companies, the shift forces them to compete on more than just technical excellence. Usability, integration, and polished application layers will be where they differentiate.

What’s Next?

The divide between open-source and proprietary models is narrowing fast. Tools like DeepSeek make AI accessible to more people, while premium platforms push boundaries in usability and polish.

For most users, though, the real question isn’t “how advanced is this model?” It’s “how well does this work for me?” That’s where the true competition will play out—in workflows, tools, and seamless integrations that make AI indispensable.

For businesses and developers, this is an incredible moment of opportunity. Whether you’re building on open-source tools or proprietary platforms, the focus should be on turning models into solutions.

By Daniel Huszár