Nvidia’s data center “real estate” is facing pressure as competitors ramp up production of AI chips. Advanced Micro Devices (AMD) is increasing production of its MI300X GPU, which is more affordable than Nvidia’s H100. Intel is also preparing to launch its Gaudi 3 AI accelerator in the second half of 2024.
While these new entrants offer certain advantages, Nvidia’s chips are expected to maintain their compute advantage. However, the inability to meet overwhelming demand may drive buyers toward AMD and Intel’s less expensive alternatives, reducing Nvidia’s share in high-compute data centers. Major tech firms like Meta Platforms, Microsoft, Alphabet, and Amazon are developing AI chips to use alongside Nvidia’s GPUs.
Microsoft’s Azure Maia 100, Alphabet’s Trillium, Amazon’s Trainium2, and Meta’s Training and Inference Accelerator are expected to complement Nvidia’s offerings, potentially reducing the need for Nvidia’s AI-GPUs in these data centers. The influx of new AI chips could diminish Nvidia’s pricing power and lead to less control over data center real estate. Most businesses lack a clear strategy to monetize their AI investments, which could further impact Nvidia’s dominance as enterprise data centers evolve.
Investors should note that competitive pressure and evolving market dynamics might impact Nvidia’s market share moving forward. It’s crucial to stay informed about the industry landscape and potential challenges Nvidia faces as the AI market continues to grow and diversify. Nvidia’s recent slump is reaching technical milestones that suggest the stock could continue to slide, according to Raymond James.
Shares of the dominant maker of chips used in artificial intelligence are down about 8% in July. Raymond James chart analyst Javed Mirza said in a July 25 note to clients that Nvidia has triggered a “mechanical sell” signal based on a moving average convergence/divergence indicator, or “MACD,” which measures price momentum. It is not the only technical signal flashing sell.
“In conjunction with price moving below an important technical level at the 50-day moving average and Volume showing early signs of selling pressure, these three early technical negatives suggest an intermediate-term (1-3 month) corrective phase is attempting to take hold,” the note said. The key level to watch is Nvidia’s 50-day moving average, currently around $118 per share, according to Mirza. “A multi-day close below important technical support at the 50-day moving average would confirm a new short-term corrective phase was underway.
This would then open the door for a filling of the gap around $94.94, or another 16.9% downside from current levels,” the note added. Nvidia has been the biggest beneficiary of the excitement around artificial intelligence. Other tech companies have been spending billions to order chips from Nvidia, which is seen as far ahead of its competitors in building high-end semiconductors.
Nvidia faces growing competitive threats
The stock is up more than 400% over the past three years. However, investors seem to be growing skeptical around AI’s immediate contribution to higher corporate profits, at least in the near term.
If that view spreads to executives at Big Tech companies such as Alphabet, it could result in new orders for Nvidia slowing down. Nvidia is set to report its second-quarter results on Aug. 28.
Global semiconductor sales are on track to hit $1 trillion as early as 2030, analysts reported at the annual Semicon West trade show in San Francisco earlier this month. Needham & Co’s Charles Shi and Gartner’s Gaurav Gupta both highlighted significant growth projections, with Gupta pegging the milestone closer to 2031 to 2032. Gupta also anticipates double-digit growth for 2024 and 2025, with semiconductor capacity expected to rise more than 50% by the end of the decade.
From AI and automotive to the rise of smart spaces and commercial drones, chips are expected to remain in high demand as the forces of autonomy, labor, power, and geopolitics shape the future. Christian Gregor Dieseldorf of SEMI warned of a critical labor shortage needed to staff the more than 70 foundries expected to go live over the next few years. Elias Eliadis from Accenture suggested that AI could play a crucial role in mitigating this issue.
He reported that fully autonomous smart factories powered by generative AI, capable of operating with minimal human intervention, are in development. Gupta expects human engagement with robots to jump to 80% over the next five years, potentially driving foundry productivity. As countries around the world race to enhance national security and economic prosperity, Dieseldorf noted that billions have already been invested in homeshoring chip manufacturing across several countries including China, Japan, India, Spain, Germany, the UK, Singapore, Taiwan, Italy, Korea, and Malaysia.
Despite the US deploying $52 billion from the US CHIPS Act to bring chip production onshore, it might be insufficient for America to significantly increase its current 10% market share. Nvidia, having crossed the $3 trillion valuation mark in June, continues to lead with more than 10% of global semiconductor revenue. The AI chip giant tops a cohort that includes Intel, Samsung, Apple, Qualcomm, SK Hynix, Broadcom, AMD, Micron, STMicroelectronics, Infineon, Texas Instruments, Sony, and NXP, which collectively made up 72% of global semiconductor revenue as of the end of last year, according to IDC’s Mario Morales.
TSMC’s chief executive officer, CC Wei, explained how Nvidia has managed to maintain high margins due to favorable pricing, enabling it to procure AI chips at $600 to $700 while selling a product for $200,000. If the current boom continues, Wei suggested this could allow suppliers to gain a larger share of the profits. In summary, the semiconductor industry is set to experience unprecedented growth driven by the increasing demand across various sectors, with AI playing a pivotal role.
However, challenges such as labor shortages and geopolitical tensions pose significant hurdles that need to be addressed to sustain this growth trajectory.
