Nvidia (NASDAQ: NVDA) has seen its stock soar by 145% over the past year, thanks to its dominant role in the artificial intelligence (AI) revolution.
The company’s graphics processing units (GPUs) are the gold standard for powering AI workloads, from training complex machine learning models to running sophisticated AI applications.
Nvidia’s chips are at the heart of most advanced AI systems, commanding an estimated 80% to 95% of the AI chip market, according to various analysts.
But a recent move by Amazon (NASDAQ: AMZN) has raised questions about Nvidia’s future grip on the industry.
Amazon’s Strategic Pivot
Amazon CEO Andy Jassy recently emphasized the importance of diversity in AI chip options, cautioning that no single chip will meet all use cases.
Drawing a parallel to past trends in machine learning frameworks, Jassy suggested that just as TensorFlow wasn’t the only game in town, Nvidia’s GPUs may not always be the default choice for AI workloads.
Nvidia’s GPUs, particularly the H200, are renowned for setting benchmarks in AI performance.
However, with prices reaching up to $40,000 per unit and millions for full server racks, these high-end chips aren’t a one-size-fits-all solution.
To address the growing demand for more affordable and efficient alternatives, Amazon has been investing in its custom AI silicon—Trainium for training and Inferentia for inference.
Amazon’s approach is grounded in its previous success with custom chips.
The company’s Graviton processors, which are based on Arm architecture, have already proven to deliver 30% to 40% better price performance compared to leading x86 processors from AMD and Intel.
The aim with Trainium and Inferentia isn’t to outpace Nvidia in sheer power but to offer a more cost-effective option for those willing to trade some speed for savings.
A Broader Industry Trend
Amazon isn’t alone in this venture.
Other tech giants like Google and Meta have also turned to custom silicon, with Broadcom developing bespoke AI chips for them.
This trend is expected to grow as companies seek to optimize for both cost and energy efficiency in their AI operations.
Joseph Moore of Morgan Stanley predicts that demand for custom chips will eventually surpass that for general-purpose GPUs.
As hyperscalers like Amazon, Google, and Meta continue to scale their AI infrastructure, the market is likely to see a shift toward more energy-efficient, cost-optimized solutions.
The Future of Nvidia
So, where does this leave Nvidia? While the company’s market dominance isn’t under immediate threat, it’s clear that competition is heating up.
Amazon’s move, backed by broader industry trends, suggests Nvidia might lose some market share in the coming years.
Moreover, the company could face pressure on pricing as alternatives emerge.
Indeed, Nvidia’s gross margin dropped 330 basis points last quarter, even as it posted 122% revenue growth.
This trend could persist, signaling that the days of sky-high margins might be numbered. However, Nvidia remains a strong player in the AI space.
Wall Street still expects the company’s earnings to grow at an impressive 37% annually over the next three years, making its current valuation appear reasonable.
For investors, Nvidia remains a key player in the AI boom.
As Jim Kelleher of Argus puts it, most technology investors should have a position in Nvidia, especially “in the age of deep learning, AI, and GPU-driven application acceleration.”
In summary, while Amazon’s strategic shift is a wake-up call, Nvidia’s stronghold on the AI chip market is far from over.
But the road ahead might be bumpier than the past year’s meteoric rise suggests, particularly as the business community continues to explore more cost-effective solutions.
You May Also Like: Amazon and Alphabet Poised to Outshine Nvidia by 2029: A New Era in AI
