Morgan Stanley has reiterated its bullish NVIDIA outlook following the company’s Computex 2026 presentation in Taiwan. The bank maintained its Overweight rating and $288 price target while naming NVIDIA its top processor stock.
Analyst Joseph Moore and his team cited NVIDIA’s GPU leadership and growing CPU business. They also identified new growth opportunities across AI servers, data centres and personal computers.
“We attended NVIDIA’s Computex keynote as well as an analyst Q&A in Taiwan this week, and came away more bullish,” Moore and his team wrote. Meetings with management increased their confidence in NVIDIA’s position across graphics processors and CPUs.
Morgan Stanley called NVIDIA the best value among the processor companies under its coverage. Analysts also said NVIDIA holds a strong position across chips, networking, software and complete computing systems.
Computex Taipei brings together technology companies, manufacturers and industry executives each year. NVIDIA used the 2026 event to outline its latest plans for AI infrastructure and personal computing.
Management expressed confidence in both client and server CPUs, according to Morgan Stanley. These products could create new revenue sources beyond NVIDIA’s established graphics processor business.
NVIDIA confirmed that its Vera Rubin platform had entered production. The system will succeed Blackwell and support large AI models, inference workloads and advanced computing systems.
Morgan Stanley expects Rubin to support NVIDIA’s next product cycle. Customer demand, production schedules and spending by major cloud providers will shape its commercial performance.
Meanwhile, NVIDIA introduced RTX Spark, a processor built for AI-focused Windows computers. The chip combines central and graphics processing functions for machines designed to run AI agents and related tools.
Computer manufacturers plan to launch systems using RTX Spark. NVIDIA will compete with Intel, AMD and Qualcomm as it expands further into the personal computer market.
Morgan Stanley estimates that NVIDIA controls about 85% of the AI accelerator market. Analysts also said they have seen no clear market share loss during the past two years.
NVIDIA focused part of its presentation on the total cost of operating AI inference systems. Its approach combines processors, networking equipment, software and server systems within one platform.
Management addressed time to first token, output per watt, reliability and the useful life of installed equipment. These measures can affect the operating cost and performance of large AI data centers. “They are all things that the customers can only figure out after deploying systems at scale, not in benchmarks or lab testing,” Moore and his team wrote. The analysts said customers need large-scale deployments to assess those operating measures accurately.
NVIDIA says its complete platform can provide better performance for each dollar spent. Actual customer deployments will determine how its systems perform across different AI workloads.
Morgan Stanley issued its NVIDIA update after reviewing Broadcom’s fiscal second-quarter results. Broadcom reported revenue of $22.19 billion, up 48% from the same quarter last year.
AI semiconductor revenue reached $10.8 billion and increased 143% year over year. Broadcom shares fell as investors compared the results and outlook with high market expectations.
Moore called the post-earnings decline ‘somewhat surprising.’ Morgan Stanley maintained its Overweight rating on Broadcom and said the company continues to record strong AI growth.
Moore said NVIDIA trades at a lower valuation and receives less market credit for the Rubin product cycle. He kept NVIDIA as Morgan Stanley’s top pick while ranking Broadcom above most other computing companies under the firm’s coverage.
Also Read: NVIDIA Bets on AI PCs With New RTX Spark Windows Laptops and Desktops