The ups and downs in NVIDIA’s price
NVIDIA (NVDA) has been on a roller-coaster ride since October due to a variety of company- and industry-specific factors and a weak macroeconomic environment. In October, crypto-related demand vanished, creating a glut of secondhand GPUs (graphics processing unit) in the market. At the same time, the US imposed tariffs on $200 billion worth of Chinese exports. Amid weak demand, NVIDIA launched the industry’s first real-time ray tracing gaming GPUs based on its Turing architecture. The GPUs failed to create a spark for a few reasons:
- their high prices
- the absence of a ray-tracing-supported ecosystem
- the marginal performance improvement they provided to existing AAA games
These headwinds pulled NVIDIA stock down as much as 55% in the fourth quarter. The company saw some growth in March and April followed by declines when the US-China trade war reignited in May. Higher tariffs have affected its earnings guidance for the rest of the year. Signs of the trade war’s easing brought some relief for the stock in July.
NVIDIA is desperate to boost its gaming revenue
NVIDIA is a leader in the discrete GPU market and earns more than 50% of its revenue from gaming. However, its gaming revenue has fallen in the high double digits in the last two quarters. Its fiscal 2020 first-quarter gaming revenue fell 39% YoY (year-over-year) as it moved to clear GPU inventory after the crypto bubble burst. However, its gaming revenue rose 10.6% sequentially due to its ray-tracing-supported GTX 1600 Series GPU starting at $149.
The two consecutive quarters of YoY declines in gaming revenue reflected in NVIDIA’s discrete GPU market share. It fell from 81.2% in the fourth quarter of 2018 to 77.3% in the first quarter of 2019. NVIDIA lost this market share to Advanced Micro Devices (AMD).
Falling gaming revenue and lost market share have pressured NVIDIA to boost its gaming business and keep its lead. NVIDIA’s desperation was visible in its Turing GPU line-up, which was a bit messy. It launched too many gaming GPUs in too short a time frame to cover every price range.
NVIDIA’s Turing GPU portfolio
NVIDIA’s Turing GPU line-up has ten GPUs compared to four in the Pascal line-up. NVIDIA started with the Turing launch with the standard versions of the RTX 2080 Ti, 2080, 2070, and 2060. It followed this traditional GeForce GPU range in the Maxwell and Pascal generations as well.
The Turing GPUs brought significant improvements in ray tracing performance, but it delivered similar performances to its Pascal predecessor in games that didn’t support real-time ray tracing. The lack of ray-tracing-supported games and the high price of RTX GPUs discouraged gamers from upgrading.
To boost Turing GPU adoption, NVIDIA launched the Turing-based GTX 1660 Ti, 1660, and 1650 without ray tracing for $149–$279. These GPUs offered up to 50% performance improvements over their Pascal-based predecessors. These GPUs increased NVIDIA’s gaming revenue in the first quarter.
In July, AMD launched its Navi-based RX 5700 Series GPUs. It claims that these GPUs beat NVIDIA’s RTX 2070 in gaming performance (except for ray tracing). To compete with AMD’s new GPU line-up in price and performance, NVIDIA is launching a refreshed version under the Super GPU line-up at the same price point as its standard version.
NVIDIA Super GPU line-up
NVIDIA unveiled the Super version of its RTX 2060, 2070, and 2080 GPUs on July 2, just a few days before the launch of AMD’s Navi GPUs. NVIDIA improved the product specifications in its refreshed series by adding more cores, boosting the clock speed, and improving the Tera floating-point operations per second performance. The Super line-up will offer an average 15% speed boost over the standard versions, retaining NVIDIA’s GPU performance advantage.
The specifications of the 2070 and 2060 Super cards show that NVIDIA is taking a hit on its margins by offering these specs at the same price points as its standard versions. For instance, NVIDIA moved to a bigger chip for the RTX 2070 Super to fit in extra transistors for the same price and increased the video random-access memory for the RTX 2060 Super to 8 GB compared to the RTX 2060’s 6 GB for just $50 more.
Although NVIDIA improved the specs of its RTX Super GPUs, it compromised on its TDP (thermal design power) advantage. Until now, NVIDIA’s GPUs had a significant TDP advantage over AMD’s GPUs. With the RTX Super GPU line-up, NVIDIA increased its TDP compared to the standard versions, making it less power efficient and reducing the gap between NVIDIA’s and AMD’s GPUs.
NVIDIA’s latest launch
NVIDIA will launch its RTX 2060 and 2070 Super GPUs on July 9, two days after the launch of AMD’s RX 5700 Series GPUs. NVIDIA stated that the RTX 2070 Super and 2080 Super will replace their standard versions. It will pack the new Super GPUs with two free ray-tracing-supported games: Remedy Entertainment’s Control and Bethesda’s Wolfenstein: Youngblood.
As the Super GPU line-up hits the market, retailers will automatically reduce the prices of the standard version GPUs. The RTX 2080 Ti will continue to be the premium offering, with the RTX 2060 being the lower-end option.
NVIDIA wanted to put pressure on AMD with its Super GPU line-up, but the latter played its hand and reduced the prices of its Navi-based GPUs. AMD was able to cut its prices because of the cost advantage offered by the 7 nm (nanometer) node. With its Navi GPUs, AMD looks to gain discrete GPU market share. Now NVIDIA is looking to get into the 7 nm game.
NVIDIA to switch to Samsung for 7 nm gaming GPUs
Over the last year, NVIDIA has been bombarded with questions about when it will transition from the 12 nm node to the 7 nm node, as its rival AMD overtook it in node technology by transitioning. NVIDIA stated that it wasn’t in a rush to transition to 7 nm GPUs, as its Turing GPUs are more energy efficient than AMD’s 7 nm GPUs.
However, a report by DigiTimes on June 5 stated that NVIDIA Korea had said it would use Samsung’s (SSNLF) 7 nm extreme ultraviolet node for its new GPUs—likely for the full range. Until now, NVIDIA manufactured most of its GPUs on TSMC’s (TSM) nodes. It last used Samsung’s foundry for its Pascal-based GTX 1050, 1050 TI, and GT 1030 cards. Last year, AMD switched to TSMC for its complete range of 7 nm products. This may be why NVIDIA is switching to Samsung.
As Samsung struggles to overtake TSMC in the 7 nm node space, it’s reportedly offering good discounts to attract customers. This foundry competition will help NVIDIA produce its new GPUs for less, and it could pass the benefits to customers.
NVIDA’s 7 nm gaming GPU to come next year
NVIDIA has already launched a refreshed version of its Turing-based GPUs on the 12 nm node. Citing market sources, DigiTimes reported that NVIDIA’s 7 nm GPUs will be built on its next-generation Ampere architecture. Samsung will reportedly start sampling its 7 nm node with customers such as NVIDIA by the end of the year, and it’s set to begin volume production by mid-2020. NVIDIA launched its Pascal series in mid-2016. Looking at Samsung’s and NVIDIA’s production and launch schedule, we can expect NVIDIA’s 7 nm GPUs to launch in mid-2020.
Next year will be exciting for the GPU market. AMD will launch its Navi 20 GPUs, NVIDIA will launch its first 7 nm GPU, and Intel will launch its first discrete GPU Xe.
Risks in NVIDIA’s gaming business
Until now, NVIDIA has been unbeatable in the GPU space. This positioning has given it the power to command a higher price for its products. However, the scenario is changing.
NVIDIA’s GPUs take the top ten spots on the list of the most preferred GPUs by users of Steam, a digital platform that allows users to install and update games. NVIDIA’s Pascal-based GTX 1060 is the most widely used GPU by Steam users. AMD’s Polaris-based RX 580 GPU is in 11th place. The Steam data shows that PC gamers aren’t upgrading their Pascal-based GPUs.
Gamers are likely sticking to their old cards because of a lack of ray-tracing-supported games. NVIDIA’s move to unlock Turing’s ray tracing feature in Pascal cards didn’t boost the adoption of Turing. AMD is preparing to bring ray tracing to Sony’s and Microsoft’s next-generation consoles, which means NVIDIA will lose its technology advantage. The adoption of cloud gaming could also slow the adoption of GPUs by gamers, but it will open a new revenue opportunity in subscriptions.
All of the above factors have affected GPU demand. According to Jon Peddie Research data, overall GPU shipments fell 18.6% sequentially in the first quarter.
What lies ahead?
With its strong balance sheet and profit margins, NVIDIA could indulge in a price war with AMD, but that would hurt its margins. NVIDIA’s high margins from the gaming and data center businesses pay for its research and development related to future technologies, such as AI and autonomous driving. Until these future markets mature and start generating adequate revenue, NVIDIA has to protect its gaming and data center margins and market share by taking calculated risks.