NVIDIA CEO Jensen Hwang speaks at a press conference after arriving at Gimpo International Airport in Seoul, South Korea on June 5, 2026.
Chris John | Null Photo | Getty Images
Nvidia’s stock price has been depressed recently, and Kalsi’s traders predict that the amount the company can charge for its chips is also falling.
While the artificial intelligence chip maker is up about 12% in 2026, its stock price has plummeted over the past month, down about 3%. In comparison, VanEck Semiconductor ETF (SMH) It’s up 84% this year, and the fund is up 15% over the past month.
As Wall Street focuses on memory chips and infrastructure as the next step in building AI, NVIDIA is sitting on the sidelines. This has resulted in benefits for companies such as: micron technology and sandiskboth of which are up nearly 60% in the past month alone.
As Nvidia’s stock price has slumped, the price of its B200 chip’s computing power has also fallen. Nvidia’s B200 is the company’s flagship graphics processing unit, or GPU, that helps run large data centers.
According to Ornn, which provides live GPU compute prices for major hardware types, B200 hourly compute prices rose to $6.11 on May 30, the highest price in three months. Since that date, the dashboard shows that the price per hour for B200 compute has fallen, to $4.22 as of June 21st.
Karshi traders are now pessimistic that the computing price of Nvidia’s AI chips will rise above May’s highs. The deal calls for pricing for Nvidia B200 computing in the second quarter and will be resolved if the price shown on Ornn crosses a certain threshold by June 30th.
Most companies rent access to GPUs through cloud providers or Neocloud, another growing ecosystem. However, GPU rental costs can fluctuate as demand for AI infrastructure grows.
“Many people don’t know how much computing power they’re going to need next year, and many suppliers of that computing power currently don’t know how many GPUs to order and at what capacity,” Soyoung Kim, a finance professor at Santa Clara University, previously told CNBC. “And manufacturers like Nvidia don’t know how much to produce.”
Earlier this month, Google agreed to pay SpaceX $920 million a month to rent AI computing power from October 2026 to June 2029. In doing so, Google will use approximately 110,000 Nvidia GPUs, CPUs, memory, and other related components. Following the deal, RBC Capital Markets was bullish on Nvidia’s performance in the second half of 2026 and 2027, saying the semiconductor giant “appears to be best positioned among its peers.”
“Regardless of the exact rationale, these GPU rental deals should allay any deep-seated concerns that NVDA will lose market share to (application-specific integrated circuits), at least in the short term,” the analysts wrote.
—CNBC’s Yun Li contributed to this article.
Disclosure: CNBC and Kalsi have a commercial relationship that includes customer acquisition and minority ownership.
