How Liang Wenfeng Made DeepSeek for 93% Less Than OpenAI

In partnership with

Introduction

What if you could achieve groundbreaking results without needing massive resources? In the AI world, where companies like OpenAI reportedly spent between $80 million and $100 million to train GPT-4, it’s easy to assume that innovation requires deep pockets. But then there’s DeepSeek, a Chinese AI startup led by founder Liang Wenfeng, which “came out of nowhere” last week, with claims to have trained its advanced model for just $5.6 million (~7% of the cost of their US counterparts).

This stark contrast raises an important question: What if success isn’t about how much money you spend—but how creatively you use what you have?

In this article, we’ll explore two key lessons from Liang Wenfeng’s journey that every micropreneur can apply to their own ventures: Resourcefulness Over Resources and Challenging Assumptions About Scale. These principles will hopefully inspire you to think differently, innovate efficiently, and build your “micro empire” even with limited means.

You Don’t Need a Billion-Dollar Budget to Innovate

Most people assume that cutting-edge AI development demands vast sums of money. For instance, OpenAI’s soaring costs are projected to reach $7 billion for 2024 alone, while other tech giants invest billions into infrastructure and compute power. Yet, Liang Wenfeng proved that resourcefulness—not raw spending—is the true driver of innovation.

DeepSeek achieved state-of-the-art performance not by outspending competitors but by optimizing every dollar. The company focused on lean operations, efficient algorithms, and maximizing existing tools. Instead of relying on extravagant budgets, they prioritized high-impact decisions that aligned with their goals.

For example, rather than building everything from scratch, DeepSeek likely utilized open-source frameworks and pre-trained models to reduce costs significantly. This approach allowed them to deliver exceptional results without breaking the bank.

At MicroEmpires, we’ve shared our own journey of using AI as a powerful ally in content creation. In Issue #128, we explain how to harness AI efficiently—without letting ‘the machines’ take over. Similarly, micropreneurs can adopt cost-effective strategies like leveraging pre-trained datasets or cloud services with pay-as-you-go pricing.

Key Takeaway for Micropreneurs: Innovation isn’t about spending more; it’s about spending smarter. Focus on projects that deliver tangible value instead of trying to do everything at once.

Call to Action: Audit your current workflows today. Where can you cut waste or repurpose existing assets? Small changes can lead to big wins.

Bigger Isn’t Always Better—Smarter Is Better

Another common misconception is that bigger teams, larger datasets, or more compute power automatically lead to better outcomes. However, Liang Wenfeng challenged this assumption head-on. By focusing on efficiency and targeted problem-solving, DeepSeek demonstrated that smaller-scale solutions often outperform bloated ones.

Reports suggest that DeepSeek’s training process relied on approximately 2,000 GPUs—a fraction of what many competitors use. Despite this, the company managed to produce highly competitive AI models. Their success shows that scaling up doesn’t always mean improving—it’s about finding the right balance between capability and necessity.

Moreover, some analyses argue that DeepSeek’s reported $5.6 million training cost may be misleading when considering broader infrastructure investments. Even so, the core lesson remains: Liang Wenfeng’s team avoided unnecessary complexity and stayed hyper-focused on delivering practical value.

As we’ve discussed in our guide to understanding micro businesses, staying lean and adaptable is crucial for long-term success. Micropreneurs should focus on curating smaller, high-quality datasets tailored to their specific use case instead of collecting massive amounts of generic data.

Key Takeaway for Micropreneurs: Question what “scale” really means for your business. Do you truly need massive datasets or endless features, or can you achieve similar results with simpler, streamlined solutions? Sometimes, less is more.

Call to Action: Reflect on your own assumptions about growth and scale. Are you chasing size for its own sake, or are you prioritizing meaningful impact?

The Power of Constraints

Liang Wenfeng’s story underscores a powerful truth: Constraints can fuel creativity and innovation. By embracing limitations, he turned perceived weaknesses into strengths, proving that extraordinary results are possible even with limited resources.

As a micropreneur, you have a unique advantage—you’re nimble, adaptable, and unafraid to think differently. With the right mindset, you can achieve remarkable outcomes without needing a billion-dollar budget.

So, here’s my final challenge to you: What’s one area of your business where you can apply these principles today?

References:

1. OpenAI's Costs Soar to $7B by 2024

2. DeepSeek Debates: Chinese Leadership On Cost, True Training 

3. Building a Social Media Empire: Leveraging AI for Unparalleled 

4. How We Use AI to Create Micro Empires Content 

5. Understanding micro business: Key insights | Wolters Kluwer

6. "DeepSeek $6M Cost Of Training Is Misleading"