Nvidia Impresses At 2017 Supercomputing Conference, KeyBanc Says

After NVIDIA Corporation (NASDAQ: NVDA) reported stellar quarterly results, which propelled the stock by over 5 percent, the company is back in the spotlight yet again.

Nvidia hosted two investor events at the 2017 Supercomputing Conference, or SC17, on Monday.

The Analyst

KeyBanc Capital Markets analyst Michael McConnell maintained his Sector Weight rating, basing his neutral view primarily on valuation. The analyst's fair value estimate for Nvidia is $200, or 40 times his 2019 earnings per share estimate of $5.

The Thesis

Strong GPU demand for cryptocurrency mining has exhausted excess inventory from the channel, triggering strong OEM sales, McConnell said in a Monday note. (See McConnell's track record here.)

This has offset concerns about gaming segment growth for the remainder of the year, McConnell said. 

See also: Game On: Intel Is Targeting Nvidia And AMD's Market Share

McConnell's takeaways from the Supercomputing Conference were:

• Nvidia reiterated its total addressable market, or TAM, estimate for datacenter at $30 billion, with the break-up among high performance computing, deep learning training and deep learning inference at $4 billion, $11 billion and $15 billion, respectively.

• Nvidia estimates that one Volta GPU node can replace 40 CPU-only servers. Nvidia's recently launched DGX-1 system with eight Volta GPUs could replace 400 CPU-based servers.

• The company is of the view that customer switching costs to FPGAs or x86 accelerators are prohibitive.

• Nvidia expects high-teens percentage growth in operating expenses this fiscal year to modestly decline in fiscal year 2019. This, according to the company, should provide more earnings leverage.

The Price Action

Nvidia shares have almost doubled year-to-date. After climbing 5 percent last Friday in reaction to the quarterly results, the stock pulled back by about 1.6 percent Monday.

Related Link:

Expect Nvidia, AMD To Ride Cryptocurrency Wave Into 2018

Photo courtesy of Nvidia.