DeepSeek’s New AI Model Tests Nvidia’s $5 Trillion Rally as Cheaper Chinese Challenger Enters the Race

Nvidia has just crossed a historic milestone, becoming the first publicly traded company to reach a market capitalisation above $5 trillion. But even as investors celebrated the record breaking rally, a fresh challenge emerged from China that could reshape the conversation around artificial intelligence spending, chip demand, and valuation expectations.

Chinese AI company DeepSeek unveiled its new V4 Pro large language model on the same day Nvidia ended trading with a market value of about $5.06 trillion. The timing was striking. While Nvidia’s surge reflected investor confidence in its leadership of the AI boom, DeepSeek’s latest release revived questions about whether cheaper and more efficient AI systems can reduce dependence on expensive computing hardware.

Nvidia has become one of the biggest winners of the global AI race. Its graphics processing units power many of the world’s most advanced AI systems, and demand from cloud providers, research labs, and corporations has pushed revenue and profits sharply higher.

The company’s rise has also been driven by a widespread belief that cutting edge AI requires massive computing power and premium chips. That assumption helped Nvidia dominate the market and justify its extraordinary valuation.

However, DeepSeek’s latest announcement suggests the competitive landscape may be changing faster than expected.

According to details shared in the report, DeepSeek’s V4 Pro uses a 1.6 trillion parameter architecture. Yet only 49 billion parameters are activated per token, a design aimed at delivering frontier level performance with far lower computing costs.

That efficiency claim is central to DeepSeek’s pitch. If accurate, it means users may be able to access high end AI capabilities without the massive infrastructure spending usually associated with top tier models.

The pricing comparison has attracted particular attention. DeepSeek V4 Pro reportedly costs about $3.48 per million output tokens. That compares with around $30 for Anthropic Claude Opus and $25 for OpenAI models for similar workloads.

A lower priced version called V4 Flash is reportedly available for just $0.28 per million tokens, making it one of the cheapest advanced AI offerings in the market.

Nvidia’s premium valuation depends heavily on the idea that demand for powerful chips will continue accelerating for years. If companies can build competitive AI products with fewer resources, lower operating costs, or alternative hardware, investors may begin reassessing long term spending forecasts.

This would not necessarily weaken Nvidia overnight. The company remains deeply embedded in global AI infrastructure, with unmatched software tools, ecosystem advantages, and broad customer relationships.

But DeepSeek’s emergence adds pressure to a key market assumption: that the most advanced AI always requires the most expensive hardware.

Another notable element in DeepSeek’s latest model is the reported use of chips from Huawei Technologies rather than Nvidia hardware.

That detail is significant because it suggests China’s domestic semiconductor ecosystem may be progressing despite export restrictions and supply chain pressure. If Chinese AI developers can train and serve powerful models using local chips, it could gradually reduce reliance on US technology.

For global markets, this introduces a second challenge for Nvidia beyond pricing competition: the rise of alternative hardware ecosystems.

Markets have seen this pattern before. In early 2025, Nvidia lost nearly $600 billion in market value after DeepSeek’s earlier R1 model sparked fears that AI development costs could fall sharply.

At the time, investors worried that cheaper AI models might slow spending on expensive data centre infrastructure. Nvidia later recovered strongly and has since climbed to new highs, showing how resilient confidence in the company remains.

Still, that episode demonstrated how sensitive Nvidia’s stock can be to signs that the economics of AI are shifting.

Many analysts would argue yes. Nvidia still leads in AI chips, software frameworks, developer adoption, and enterprise trust. Major cloud providers continue investing billions in AI data centres, and demand for high performance compute remains strong.

Even if cheaper models gain traction, AI usage itself may expand faster because lower costs allow more businesses to adopt advanced tools. In that scenario, Nvidia could still benefit from broader market growth.

However, expectations are now extremely high. At a $5 trillion valuation, investors may expect Nvidia to continue outperforming quarter after quarter. Any slowdown in growth, weaker margins, or signs of stronger competition could invite volatility.

DeepSeek’s rise highlights a broader shift in artificial intelligence. The next phase of competition may not be won only by raw model size or giant spending budgets. Efficiency, affordability, and deployment flexibility are becoming just as important.

That could benefit startups, enterprises, and countries seeking lower cost alternatives to dominant US platforms. It may also intensify competition among model developers such as OpenAI, Anthropic, Google, Meta, and Chinese challengers.

For users, lower prices and stronger competition could accelerate innovation and make advanced AI more widely accessible.

DeepSeek’s V4 Pro may not immediately derail Nvidia’s $5 trillion celebration, but it arrives at a moment when markets are watching every signal in the AI race. Nvidia remains the clear leader in hardware, yet DeepSeek is challenging the belief that frontier AI must always come with frontier level costs.

That debate matters far beyond one stock price. It could define how the next generation of artificial intelligence is built, priced, and distributed around the world.

For now, Nvidia still leads the party. But DeepSeek has made sure investors know the competition has arrived.