We use cookies to understand how you use our site and to improve your experience. This includes personalizing content and advertising. To learn more, click here. By continuing to use our site, you accept our use of cookies, revised Privacy Policy and Terms of Service.
You are being directed to ZacksTrade, a division of LBMZ Securities and licensed broker-dealer. ZacksTrade and Zacks.com are separate companies. The web link between the two companies is not a solicitation or offer to invest in a particular security or type of security. ZacksTrade does not endorse or adopt any particular investment strategy, any analyst opinion/rating/report or any approach to evaluating individual securities.
If you wish to go to ZacksTrade, click OK. If you do not, click Cancel.
On Tuesday I was a guest on Tracey Ryniec's MarketEdge podcast discussing the opportunities and the risks for investors in semiconductor stocks.
After we covered the basics involving memory chip and equipment makers like Micron and Lam Research, we talked about the "celebrity" of NVIDIA (NVDA - Free Report) .
I proposed that the company's stature and valuation at nearly 40X is deserved because it thrives in a unique universe of new AI growth industries and technologies that it helped create.
As I defended the stock price and valuation, Tracey eventually asked something like, "Okay, so what dethrones the king then?"
My answer was just two things: either Intel (INTC - Free Report) or Alphabet (GOOGL - Free Report) launch a competing platform for machine learning and deep learning hardware customers like Microsoft (MSFT - Free Report) , Amazon (AMZN - Free Report) and other industrial, medical, automotive and university research labs.
Probably within the hour of our discussion being finished, Google made a big, scary reveal for NVDA investors with the launch of TPU 3.0 (Tensor Processing Unit), their next-generation of machine learning processors.
FLOPS, Teraflops, and Petaflops
In the video that accompanies this article I explain with some very pretty pictures what exactly Google has been up to. Basically, the monster of the web and big data not only created their own programming language for AI applications called Tensor, they've been busy designing and building the chips to run it for the past three years.
I also explain some of the processing capacity of these monster machines, with stacks of billions of transistors. For comparison's sake, Microsoft's Xbox One has 6 teraflops, or the ability to do 6 trillion "floating-point operations per second" (thus, "flop").
Google's TPU 3.0 ramped up the power of their old version from May of last year by 8 times -- and 2.0 was already at 11.5 petaflops! Yes, "peta" is 10 to the 15th power, so 11.5 quadrillion flops.
They achieved that 2017 mark by joining together 256 chips. With 3.0, I imagine we're talking about over 2,000 chips and capacity approaching 100 petaflops.
Check out my video for a great visual on the "massively parallel architecture" that was TPU 2.0. I can barely imagine how many square feet 3.0 takes up (wait till you hear about the cooling technology).
Citron Reiterates Their NVDA Short Thesis
Since NVIDIA reports earnings on Thursday May 10, Citron Research jumped on the Google news as just what the doctor ordered for NVDA's rich valuation. Here was their post on Twitter Tuesday...
"Not a story of shorts vs $NVDA, it is $NVDA vs $GOOG: TPU 3.0 announced today is 8x faster than 2.0 and liq cooled. This happens when customer competes. We expect the press to write soon. Shorting into earnings. Good company dangerous stock price."
Citron included a link with a photo in the tweet that showed another radical innovation from Google: liquid cooled processors. You can get a closer look in my video.
The reason that Citron should be taken seriously here is that they understand the markets that NVIDIA and Google serve with these platforms.
And Google has outdone itself in perfecting its industrial-grade AI platform TensorFlow and inviting big data users to lease cloud space from them to do their heavy "crunching." No need to buy any equipment when you can just rent it, and from anywhere.
So, it's perfectly logical that Google would seek to master both the hardware and the software of the AI cloud. And, of course, NVDA CEO Jensen Huang and Co. knew this was coming.
Bottom line on NVDA vs. Google: Competition is good. I think it will make NVIDIA better.
You can hear more about Google's deep learning research that was harnessed to beat world champs in the incredibly complex Chinese game Go, and NVIDIA’s engine that conquered Texas Hold'em poker, in my podcast from last October...
Disclosure: I own shares of NVDA for the Zacks TAZR Trader portfolio.
Kevin Cook is a Senior Stock Strategist for Zacks Investment Research where he runs the TAZR Trader service. Click Follow Author above to receive his latest stock research and macro analysis.
The Hottest Tech Mega-Trend of All
Last year, it generated $8 billion in global revenues. By 2020, it's predicted to blast through the roof to $47 billion. Famed investor Mark Cuban says it will produce "the world's first trillionaires," but that should still leave plenty of money for regular investors who make the right trades early.
Image: Bigstock
Google Ramps Assault on NVIDIA AI Throne
On Tuesday I was a guest on Tracey Ryniec's MarketEdge podcast discussing the opportunities and the risks for investors in semiconductor stocks.
After we covered the basics involving memory chip and equipment makers like Micron and Lam Research, we talked about the "celebrity" of NVIDIA (NVDA - Free Report) .
I proposed that the company's stature and valuation at nearly 40X is deserved because it thrives in a unique universe of new AI growth industries and technologies that it helped create.
As I defended the stock price and valuation, Tracey eventually asked something like, "Okay, so what dethrones the king then?"
My answer was just two things: either Intel (INTC - Free Report) or Alphabet (GOOGL - Free Report) launch a competing platform for machine learning and deep learning hardware customers like Microsoft (MSFT - Free Report) , Amazon (AMZN - Free Report) and other industrial, medical, automotive and university research labs.
Probably within the hour of our discussion being finished, Google made a big, scary reveal for NVDA investors with the launch of TPU 3.0 (Tensor Processing Unit), their next-generation of machine learning processors.
FLOPS, Teraflops, and Petaflops
In the video that accompanies this article I explain with some very pretty pictures what exactly Google has been up to. Basically, the monster of the web and big data not only created their own programming language for AI applications called Tensor, they've been busy designing and building the chips to run it for the past three years.
I also explain some of the processing capacity of these monster machines, with stacks of billions of transistors. For comparison's sake, Microsoft's Xbox One has 6 teraflops, or the ability to do 6 trillion "floating-point operations per second" (thus, "flop").
Google's TPU 3.0 ramped up the power of their old version from May of last year by 8 times -- and 2.0 was already at 11.5 petaflops! Yes, "peta" is 10 to the 15th power, so 11.5 quadrillion flops.
They achieved that 2017 mark by joining together 256 chips. With 3.0, I imagine we're talking about over 2,000 chips and capacity approaching 100 petaflops.
Check out my video for a great visual on the "massively parallel architecture" that was TPU 2.0. I can barely imagine how many square feet 3.0 takes up (wait till you hear about the cooling technology).
Citron Reiterates Their NVDA Short Thesis
Since NVIDIA reports earnings on Thursday May 10, Citron Research jumped on the Google news as just what the doctor ordered for NVDA's rich valuation. Here was their post on Twitter Tuesday...
"Not a story of shorts vs $NVDA, it is $NVDA vs $GOOG: TPU 3.0 announced today is 8x faster than 2.0 and liq cooled. This happens when customer competes. We expect the press to write soon. Shorting into earnings. Good company dangerous stock price."
Citron included a link with a photo in the tweet that showed another radical innovation from Google: liquid cooled processors. You can get a closer look in my video.
The reason that Citron should be taken seriously here is that they understand the markets that NVIDIA and Google serve with these platforms.
And Google has outdone itself in perfecting its industrial-grade AI platform TensorFlow and inviting big data users to lease cloud space from them to do their heavy "crunching." No need to buy any equipment when you can just rent it, and from anywhere.
So, it's perfectly logical that Google would seek to master both the hardware and the software of the AI cloud. And, of course, NVDA CEO Jensen Huang and Co. knew this was coming.
Bottom line on NVDA vs. Google: Competition is good. I think it will make NVIDIA better.
You can hear more about Google's deep learning research that was harnessed to beat world champs in the incredibly complex Chinese game Go, and NVIDIA’s engine that conquered Texas Hold'em poker, in my podcast from last October...
AI in Action: DeepStack, DeepMind, and Deep Learning Intuition (Part 2)
Disclosure: I own shares of NVDA for the Zacks TAZR Trader portfolio.
Kevin Cook is a Senior Stock Strategist for Zacks Investment Research where he runs the TAZR Trader service. Click Follow Author above to receive his latest stock research and macro analysis.
The Hottest Tech Mega-Trend of All
Last year, it generated $8 billion in global revenues. By 2020, it's predicted to blast through the roof to $47 billion. Famed investor Mark Cuban says it will produce "the world's first trillionaires," but that should still leave plenty of money for regular investors who make the right trades early.
See Zacks' 3 Best Stocks to Play This Trend >>