Piper Sandler Predicts NVIDIA’s GPU Capacity to Surge by 10x and Model Inference Costs to Drop 35% Fueling the Next “AI Wave 2”

Piper Sandler Predicts NVIDIA’s GPU Capacity to Surge by 10x and Model Inference Costs to Drop 35% Fueling the Next “AI Wave 2”

This does not constitute investment advice. The author does not hold any positions in the stocks mentioned.

The Shifting Landscape of AI Investment Amid Market Turmoil

In light of the current turbulence affecting the stock market, it may seem difficult to find optimism surrounding the emerging “AI mania 2.0.”However, Piper Sandler’s Brent A. Bracelin is optimistic about this trend, pointing to NVIDIA’s remarkable expansion in GPU capabilities and a significant drop in AI model inference costs as critical drivers for this new wave of artificial intelligence investments.

Market Observations and Momentum Analysis

NVIDIA’s Stock Volatility and Industry Challenges

Recent commentary from Bernstein analyst Mark Li highlights that NVIDIA’s stock has faced significant pressures due to concerns about growth, ongoing supply chain issues, and regulatory hurdles. The company’s ability to sell GPUs, particularly in China, has been hampered by geopolitical tensions and tariff regulations. Furthermore, the implications of Trump’s tariff policies continue to stifle economic confidence, prompting a surge in recession predictions.

J. P.Morgan’s analysis of past momentum sell-offs indicates an average peak-to-trough decline of 14%.Meanwhile, Morgan Stanley’s Mike Wilson forecasts that the S&P 500 could drop to around 5, 500 in the first half of 2025, with expectations of a robust market recovery in the subsequent months.

AI Wave 2: A Bullish Outlook

Despite the prevailing market challenges, Bracelin’s latest investment report offers a more optimistic perspective regarding AI wave 2.0. He notes the astonishing tenfold increase in NVIDIA’s GPU capacity over the last two years alongside a remarkable 35-fold decrease in AI model inference expenses, largely driven by innovations from DeepSeek’s R1 and Alibaba’s Qwen AI models. These factors are poised to catalyze a resurgence in the AI sector.

To illustrate this growth, Bracelin points to NVIDIA’s data center revenue, which surged from $3.6 billion in early 2023 to $36 billion in the most recent quarter.

On the inference cost front, he highlights that OpenAI’s GPT-4o came with a $15 price point per million tokens at an efficiency of 128K tokens per second. In contrast, Alibaba’s Qwen QwQ-32B model on Groq is now priced significantly lower at $0.39 per million output tokens, achieving a performance of 400K tokens per second—demonstrating an impressive ~35-fold reduction in costs.

Potential for a New Era of AI Monetization

Bracelin identifies these developments as potential catalysts for widespread adoption of AI applications across various sectors. He asserts that the landscape appears ripe for a new wave of monetization opportunities, projecting long-term benefits to unfold over the next decade.

“We are encouraged by early proof points including 1) Microsoft Copilot having approached the $1B run-rate milestone, 2) Salesforce Data + Agentforce AI run-rate of $0.9B, 3) HubSpot with 35% of support tickets handled by Agentic AI, 4) ServiceNow with 1, 000+ customers purchasing the GenAI Pro+ WeU, and 5) Snowflake Snowpark reaching 3% of sales on >100% y/y growth.”

Conclusion: Strategic Picks for 2025

Acknowledging the recent downturn in the market, where the software sector (IGV) has seen a 12% decline compared to a 5% drop in the S&P 500, Bracelin remains hopeful that the current weakness is only temporary. He believes the forthcoming AI monetization wave will allow for a gradual recovery and a return to a long-term growth trend.

It’s thus not surprising that Bracelin highlights Salesforce (CRM), Microsoft (MSFT), and Snowflake (SNOW) as his top investment recommendations for 2025.

Source & Images

Leave a Reply

Your email address will not be published. Required fields are marked *