From NVIDIA’s latest 10-K filing:

” We do not manufacture the silicon wafers used for our GPUs and Tegra processors and do not own or operate a wafer fabrication facility. Instead, we are dependent on industry-leading foundries, such as Taiwan Semiconductor Manufacturing Company Limited and Samsung Electronics Co. Ltd., to manufacture our semiconductor wafers using their fabrication equipment and techniques. Similarly, we do not assemble, test or package our products, but instead rely on independent subcontractors, such as Advanced Semiconductor Engineering, Inc., BYD Auto Co., Ltd., Hon Hai Precision Industry Co., Ltd., JSI Logistics, Ltd., King Yuan Electronics Co. and Siliconware Precision Industries Co. Ltd. We do not have long-term commitment contracts with these foundries or subcontractors.
As a result, we face several significant risks which could have an adverse effect on our ability to meet customer demand and/or negatively impact our business operations, gross margin, revenue and/or financial results, including:

• a lack of guaranteed supply of wafers and other components and potential higher wafer and component prices due to supply constraints.”

The relevant parts are:

“We do not have long-term commitment contracts…”

and

“…we face several significant risks … including … a lack of guaranteed supply of wafers and other components…”

Therefore if NVIDIA was on the hook for Tegra TX1 production it would show up in their inventories. But it doesn’t. I’ve compiled the inventories lines of their last 7 quarters, which goes back to the 3rd quarter of calendar year 2014, before the TX1 was in volume production. All figures are in millions of US dollars, and the date listed is the closing date of a quarter.

Oct 26, 2014: 408
Jan 25, 2015: 483
Apr 26, 2015: 438
Jul 26, 2015: 441
Oct 25, 2015: 425
Jan 31, 2016: 418
May 1, 2016: 394

So, as you can see, NVIDIA’s inventories have not been building up throughout the Tegra TX1’s production. In fact, for the latest quarter they were the lowest they’ve been for the last 7 quarters.

As discussed above, there’s no “20nm deal”. And NVIDIA designed Pascal specifically for the 16nm process node. It would be a lot of work and expense to move it to 20nm. They would have to re-optimize paths and thermal characteristics, among other things. A Pascal-based Tegra would have peak single precision performance of about 1 TFLOPS on the 16nm node. This can be gleaned from NVIDIA’s Drive PX 2 information.

So you are saying that nintendo is willing to rise again after the wii u cleaning others garbage?
Seriously?!

https://www.semiwiki.com/forum/content/4300-tsmc-20nm-essentially-worthless.html

Nvidia signs agreements with TSMC to provide wafers and renting processing nodes, the contracts says if TSMC is putting money down to manufacture 20nm wafers, Nvidia will buy them. If Nvidia doesn’t keep to that agreement, TSMC will lose money.
You make a fair point on Nvidia’s desire for console contracts, it just comes with the rumor, and it can be assumed as well. There’s money to be made in console deals, Nvidia tried to re-jam their foot in the door with the Shield. It’s not doing great but not bad either, it’s just kind of there.

As for Nintendo after TX1, all of their consoles since GC have relied on old tech in a new, concise package, engineered to work together. So I wouldn’t put it past them to use a TX1 in 2017. What could be more likely is they can use the Pascal architecture on a 20nm wafer if the “20nm deal” rumor is true, as Architectures and node sizes are independent. So Nintendo and Nvidia could either work together to create a more powerful TX1, somewhere around .8-1.2Tflops or do a weird deal of using a Tegra Pascal design on a 20nm wafer, wouldn’t be the first time this happened.

the problem is anybody that is even slightly reputable could mis hear info or be fed misleading info but end of the day i highly doubt nintendo is going to release a powerful console. I think its more gimmicks that no ones going care about

well nvidia ha successfully got developers to port games to the shield console which is arm based so has to run under its own power when not connected to Grid service. its hopelessly underpowered but theyve got them to do it all the same