Bank of America equity analyst Vivek Arya continues to hold a positive outlook on Nvidia Corp.
In a note shared with clients on Thursday, Arya noted that while Nvidia's sharp rise might prompt short-term profit-taking, he believes any resulting "volatility (is) likely to be short-lived," thanks to the company's robust fundamentals and attractive valuations.
This confidence is due to several factors:
- Generative AI Hardware Deployment: We're only in the second year of what could be a 3-5 year cycle for generative AI hardware, presenting a long-term opportunity worth over $300 billion, which is three times the current year.
- New Blackwell AI Accelerator Systems: Nvidia's new AI systems, to be launched later this year, have strong demand from cloud customers.
- Growing On-Premise AI Demand: The demand for enterprise and sovereign AI, along with early stages of software monetization, is increasing.
- Attractive Valuation: Nvidia is valued at 35-40 times the consensus and about 30 times PE on a bullish earnings scenario of $5 per share, less than the expected 50%+ annual earnings growth from 2023-2026.
Arya highlighted a significant distinction between the dot-com rally and the current AI-driven surge.
According to the analyst, unlike the debt-fueled "dot-com boom," the current "genAI deployment is a mission-critical race among some of the best-funded (cloud) customers."
Bank of America maintained a Buy rating on Nvidia shares, projecting a 10.2% potential upside over the next year from Tuesday's closing price.
Growth In AI Training Requirements
Earlier this month, Nvidia's Vice President of Hyperscale and HPC Computing, Dr. Ian Buck, delivered a keynote at Bank of America's Global Tech conference. Buck highlighted Nvidia's advantage in continuously optimizing and improving its silicon, systems, and software.
He noted that large language models (LLMs), the foundation of generative AI, are rapidly growing in size, currently at around 2 trillion parameters, and doubling every six months.
Nvidia plans to build its first 100,000 GPU cluster later this year, three times the size of the largest current cluster.
Nvidia's Expanding Software Services
While Nvidia's hardware capabilities are well-known, its ability to help customers rapidly scale and deploy revenue-generating services is often underappreciated.
"We believe recurring software services could open the next leg of growth, while strengthening its direct relationship over enterprise users," Arya wrote
Nvidia's Inference Microservices (NIMs) are optimized software containers that accelerate enterprise AI application deployment from weeks to minutes.
Deploying NIMs requires an NVIDIA AI Enterprise license, starting at $4,500 per GPU per year or $1 per GPU per hour in the cloud.
- 40+ NIM Microservices are available across major models.
- 200 technology partners are integrating NIMs for domain-specific applications in sectors like manufacturing, healthcare, financial services, retail, and customer service.
- SAP SE
(SAP ) : Incorporates generative AI capabilities into SAP Joule and other cloud-native solutions. - ServiceNow Inc.
(NOW ) : Facilitates scalable, cost-effective LLM development. - CrowdStrike Holdings Inc.
(CRWD ) : Develops tailored generative AI models for business protection. - Snowflake Inc.
(SNOW ) : Integrates NeMo Retriever microservices into Snowflake's Cortex AI, linking custom models to business data. - NetApp Inc.
(NTAP ) : Introduces a converged infrastructure model for AI use without LLM training. - Dell Technologies Inc.
(DELL ) : Integrates NIMs into Dell NativeEdge for more efficient application development. - Adobe Inc.
(ADBE ) : Applies generative AI to PDF data and develops open datasets. - Synopsys Inc.
(SNPS ) : Deploys Synopsys.ai Copilot in secure, on-premise environments.
