Demand for Nvidia’s AI chip faces scrutiny after launch of low-cost DeepSeek

Demand for Nvidia’s pricey AI chips will be in focus when the company reports results on Wednesday as investors doubt the hefty spending on the technology after low-cost AI models from China’s DeepSeek rattled the industry.

The world’s second most valuable company has been the top beneficiary of an AI-driven spending spree by big technology companies over the past two years.

But claims that DeepSeek’s AI models rival its Western counterparts at a fraction of the cost have led some investors to ask if Nvidia’s cutting-edge chips are essential for gaining an edge in the AI race.

DeepSeek’s sudden rise in January resulted in Nvidia losing $593bn in market value, the largest one-day loss for any US company. Its shares were one of the best performers in 2023 and 2024.

“Investors have been very concerned about DeepSeek and the impact that it will have on demand,” said Ivana Delevska, chief investment officer of Spear Invest, which holds Nvidia shares in an actively managed exchange-traded fund.

“So if [Nvidia] can show that they’re still able to ‘beat and raise’, it would be pretty positive for the stock.”

Nvidia is expected to report a 72% surge in revenue to $38.05bn in its fourth quarter, according to LSEG data, its slowest growth in seven quarters. It is likely to forecast a 60% jump in revenue for the first quarter ending April.

In contrast, the company’s revenue has seen five straight quarters of triple-digit growth until the quarter ended October.

So far, demand for Nvidia’s AI chips has not faltered. Big customers such as Microsoft and Meta have said they plan to plow ahead with their steep data-centre spending.

“The Capex plans communicated by Meta, Microsoft, Google and Amazon … paint a very positive picture of the near-term demand backdrop for Nvidia,” said John Belton, a portfolio manager at Gabelli Funds, which holds Nvidia shares.

Meanwhile, shipments of Nvidia’s powerful Blackwell chips are expected to have accelerated in the fourth quarter, boosting its revenue but squeezing its margin due to the cost of ramping a new and complex chip.

Analysts expect Nvidia’s adjusted gross margin to shrink by more than three percentage points to 73.5% in the fourth quarter.

With its Blackwell series, Nvidia is shifting from selling individual chips to full AI computing systems such as the GB200 NVL72, which bundle GPUs, CPUs and networking equipment.

That has further complicated a costly and time-consuming production ramp-up.

Its contract manufacturer, Taiwan’s TSMC, scrambled to expand capacity for advanced packaging — a complex process that glues together chips and is the main bottleneck in AI semiconductor supply chains.

Blackwell’s rollout was also hampered by design flaws and low chip yields — although Nvidia has since fixed the issues.

In November, it said Blackwell would exceed initial revenue projections of several billion dollars in the fourth quarter.

Reuters