TLDR
- UBS raised Nvidia’s price target to $245 from $235, expecting Q4 revenue of $67.5 billion—$2.5 billion above guidance
- Broadcom’s TPUs are emerging as a cost-effective alternative to Nvidia’s GPUs, priced at $10,500-$15,000 versus $40,000-$50,000 for Blackwell chips
- UBS forecasts Broadcom will ship 3.7 million TPUs in 2026, potentially generating $60 billion in AI revenue
- Nvidia acquired a nonexclusive license for Groq’s inference technology for $20 billion to strengthen its position in the inference market
- TPUs show comparable performance to Nvidia’s GB300 chips for inference workloads but lag in training capabilities
Nvidia stock climbed 1.8% to $191.84 Tuesday as UBS raised its price target on the chipmaker ahead of earnings. The firm now sees shares reaching $245, up from a previous target of $235.
UBS analyst Timothy Arcuri expects Nvidia to report fiscal fourth-quarter revenue of approximately $67.5 billion. That figure would come in about $2.5 billion above the company’s own guidance. The firm projects first-quarter revenue of approximately $76 billion, slightly above investor expectations of $74-75 billion.
The investment bank believes Nvidia will maintain its 75% gross margin guidance. This comes despite growing questions about margin sustainability as competition heats up in the AI chip market.
Broadcom’s challenge to Nvidia’s dominance has intensified as demand grows for Google’s Tensor Processing Units. Broadcom helped design these TPUs, which are gaining traction as a more affordable alternative to Nvidia’s graphics processing units.
TPU Shipments Set to Surge
UBS forecasts Broadcom will ship around 3.7 million TPUs this year, rising to more than five million in 2027. That volume could help Broadcom generate AI revenue of around $60 billion in 2026, jumping to $106 billion in 2027.
By comparison, Nvidia is expected to generate around $300 billion in data-center sales for its fiscal 2027 year ending in January. The company’s revenue grew 65.22% over the last twelve months, cementing its position as the AI chip leader.
The pricing gap between the two chip types tells an important story. TPUs carry an average selling price of $10,500 to $15,000, with prices expected to reach $20,000 in coming years. Nvidia’s latest Blackwell chips cost between $40,000 and $50,000 per unit.
That price difference makes TPUs particularly attractive for inference workloads. Inference refers to the process of generating answers or results from AI models. Nvidia maintains an edge in training AI models, which requires more computational power.
Nvidia Moves to Strengthen Inference Position
Nvidia recently paid $20 billion for a nonexclusive license to technology from AI hardware startup Groq. The deal included compensation packages for many Groq employees who joined Nvidia. Groq specializes in inference hardware, suggesting Nvidia is making moves to protect its market position.
Benchmarks show the latest Ironwood TPU performance matches Nvidia’s GB300 for inference. However, TPU training performance is about half that of Nvidia’s chips. A model that could be trained in 35-50 days on Nvidia GPUs would take roughly three months on TPUs.
Analysts at Mizuho estimate that 20% to 40% of AI workloads currently focus on inference. That share is expected to grow to 60% to 80% over the next five years. The shift toward inference workloads could increase pressure on Nvidia’s margins.
AI startup Anthropic has placed two major TPU orders totaling $21 billion. Meta Platforms is also in talks to use the processors, according to The Wall Street Journal. These deals signal that major AI companies are exploring alternatives to Nvidia’s chips.
UBS expects Nvidia to exclude China from its guidance when it reports earnings. The country has started using more domestically produced GPUs. However, potential additional revenue from China could still boost fourth-quarter results beyond expectations.
Nvidia plans to invest over NT$40 billion to establish its first overseas headquarters in Taiwan. The Taipei city government signed a contract with Nvidia, setting land royalties at NT$12.2 billion. Apollo Global Management is nearing a $3.4 billion loan deal to purchase Nvidia chips for leasing to Elon Musk’s xAI.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
