First, Picks & Shovels… Now, an Excavator?

A couple of weeks ago, I wrote a LinkedIn post about AI’s “picks and shovels”, highlighting how infrastructure players like NVIDIA were winning big by powering the AI gold rush. Their GPUs became the backbone of AI development, making large-scale model training possible and reaping massive financial rewards in the process.

But now, a new player has entered the field**, and it’s causing some serious disruption. DeepSeek AI has burst onto the scene with an allegedly low-cost, high-performing open-source model that challenges the established AI giants. If the rumours are true, DeepSeek is proving that AI development doesn’t have to cost billions, nor does it require massive energy consumption to power it.

This is the kind of seismic shift that can reshape an industry. We all saw the immediate impact. Nvidia’s stock took a hit. And when disruption like this happens, capital doesn’t just disappear; it finds a new direction.

Where’s the capital heading?

With AI infrastructure evolving so rapidly, investors and businesses are starting to move up the stack, shifting their focus from foundational AI models to the application layer.

This means that while companies like OpenAI, Anthropic, and Google are still battling it out in the race to build bigger and better AI models, more attention is being given to how AI is actually used. The biggest value lies in AI-powered applications that directly impact businesses and users—and we’re already seeing the market react.

Just look at Salesforce. Following DeepSeek’s launch, Salesforce’s stock surged. Businesses are waking up to the reality that while AI models are important, their true power lies in turning raw data into actionable, business-changing insights.

Companies don’t just need another LLM. They need AI that is practical, usable, and integrated into their existing workflows. That’s where the real long-term value lies.

What happens next?

The rise of open-source AI models like DeepSeek raises some major questions for the industry:

• How will the major AI players adapt against lower-cost, open-source competitors?
• Could this lead to more open-source models becoming dominant, pushing the market away from closed AI ecosystems?

One thing is clear: OpenAI isn’t happy. They’ve already said they are “reviewing” allegations that DeepSeek used their AI models to train its rival chatbot.

Meanwhile, the rest of us will be watching closely… popcorn in hand.

The AI industry is at a turning point. While infrastructure has been the big winner in the AI gold rush so far, we may be entering a phase where AI-powered applications take centre stage.

Whether you’re building AI into your customer service, sales enablement, or operational workflows, the companies that embrace AI at the application level, and integrate it seamlessly into their processes, will be the ones that thrive.