Nvidia has a market cap of roughly $550 billion compared to Apple’s nearly $2.5 trillion. We believe Nvidia can surpass Apple by capitalizing on the artificial intelligence economy, which will add an estimated $15 trillion to GDP. This is compared to the mobile economy that brought us the majority of the gains in Apple, Google and Facebook, and contributes $4.4 trillion to GDP. For comparison purposes, AI contributes $2 trillion to GDP as of 2018.
While mobile was primarily consumer, and some enterprise with bring-your-own-device, artificial intelligence will touch every aspect of both industry and commerce, including consumer, enterprise, and small-to-medium sized businesses, and will do so by disrupting every vertical similar to cloud. To be more specific, AI will be similar to cloud by blazing a path that is defined by lowering costs and increasing productivity.
I have an impeccable record on Nvidia including when I stated the sell-off in 2018 was overblown and missing the bigger picture as Nvidia has two impenetrable moats: developer adoption and the GPU-powered cloud. This was when headlines were focused exclusively on Nvidia’s gaming segment and GPU sales for crypto mining.
Although Nvidia’s stock is doing very well this year, this has been a fairly contrarian stance in the past. Not only was Nvidia wearing the dunce hat in 2018, but in August of 2019, the GPU data center revenue was flat to declining sequentially for three quarters, and in fiscal Q3 2020, also declined YoY (calendar Q4 2019). We established and defended our thesis on the data center as Nvidia clawed its way back in price through China tensions, supply shortages, threats of custom silicon from Big Tech, cyclical capex spending, and on whether the Arm acquisition will be approved.
Suffice to say, three years later and Nvidia is no longer a contrarian stock as it once was during the crypto bust. Yet, the long-term durability is still being debated — — it’s a semiconductor company after all — — best to stick with software, right? Right? Not to mention, some institutions are still holding out for Intel. Imagine being the tech analyst at those funds (if they’re still employed!).
Nvidia is already the universal platform for development, but this won’t become obvious until innovation in artificial intelligence matures. Developers are programming the future of artificial intelligence applications on Nvidia because GPUs are easier and more flexible than customized TPU chips from Google or FGPA chips used by Microsoft [from Xilinx]. Meanwhile, Intel’s CPU chips will struggle to compete as artificial intelligence applications and machine learning inferencing move to the cloud. Intel is trying to catch-up but Nvidia continues to release more powerful GPUs — and cloud providers such as Amazon, Microsoft and Google cannot risk losing the competitive advantage that comes with Nvidia’s technology.
The Turing T4 GPU from Nvidia should start to show up in earnings soon, and the real-time ray-tracing RTX chips will keep gaming revenue strong when there is more adoption in 6–12 months. Nvidia is a company that has reported big earnings beats, with average upside potential of 33.35 percent to estimates in the last four quarters. Data center revenue stands at 24% and is rapidly growing. When artificial intelligence matures, you can expect data center revenue to be Nvidia’s top revenue segment. Despite the corrections we’ve seen in the technology sector, and with Nvidia stock specifically, investors who remain patient will have a sizeable return in the future.”
Notably, the stock is up 335% since my thesis was first published — a notable amount for a mega cap stock and nearly 2–3X more returns than any FAAMG in the same period. This is important because I expect this to trend to continue until Nvidia has surpassed all FAAMG valuations.
Below, we discuss the Ampere architecture and A100 GPUs, the Enterprise AI Suite and an update on the Arm acquisition. These are some of the near-term stepping stones that will help sustain Nvidia’s price in the coming year. We are also bullish on the Metaverse with Nvidia specifically but will leave that for a separate analysis in the coming month.
Nvidia Not Standing Still with Ampere Architecture and A100 GPU
“Nvidia’s acceleration may happen one or two years earlier as they are the core piece in the stack that is required for the computing power for the front-runners referenced in the graph above. There is a chance Nvidia reflects data center growth as soon as 2020–2021.” -published August 2019, Premium I/O Fund
Last year, Nvidia released the Ampere architecture and A100 GPU as an upgrade from the Volta architecture. The A100 GPUs are able to unify training and inference on a single chip, whereas in the past Nvidia’s GPUs were mainly used for training. This allows Nvidia a competitive advantage by offering both training and inferencing. The result is a 20x performance boost from a multi-instance GPU that allows many GPUs to look like one GPU. The A100 offers the largest leap in performance to date over the past 8 generations.
At the onset, the A100 was deployed by the world’s leading cloud service providers and system builders, including Alibaba cloud, Amazon Web Services, Baidu Cloud, Dell Technologies, Google Cloud platform, HPE and Microsoft Azure, among others. It is also getting adopted by several supercomputing centers, including the National Energy Research Scientific Computing Center and the Jülich Supercomputing Centre in Germany and Argonne National Laboratory.
One year later and the Ampere architecture is becoming one of the best-selling GPU architectures in the company’s history. This quarter, Microsoft Azure recently announced the availability of Azure ND A100 v4 Cloud GPU which is powered by NVIDIA A100 Tensor Core GPUs. The company claims it to be the fastest public cloud supercomputer. The news follows the launch by Amazon Web Services and Google Cloud general availability in prior quarters. The company has been extending its leadership in supercomputing. The latest top 500 list shows that Nvidia power 342 of the world’s top 500 supercomputers, including 70 percent of all new systems and eight of the top 10. This is a remarkable update from the company.
Ampere architecture-powered laptop demand has also been solid as OEM’s adopted Ampere Architecture GPUs in a record number of designs. It also features the third-generation Max-Q power optimization technology enabling ultrathin designs. The Ampere architecture product cycle for gaming has also been robust, driven by RTX’s real-time ray tracing.
In the area of GPU acceleration, Nvidia is working with Apache Spark to release Spark 3.0 run on Databricks. Apache Spark is the industry’s largest open source data analytics platform. The results are a 7x performance improvement and 90 percent cost savings in an initial test. Databricks and Google Cloud Dataproc are the first to offer Spark with GPU acceleration, which also opens up Nvidia for data analytics.
The demand has been strong for the company’s products which have exceeded supply. In the earnings call, Jensen Huang mentioned
“And so I would expect that we will see a supply-constrained environment for the vast majority of next year is my guess at the moment.” However, he assured that they have secured enough supplies to meet the growth plans for the second half of this year when he said, “We expect to be able to achieve our Company’s growth plans for next year.”
Virtual Machines for AI Workloads
Virtualization allows companies to use software to expand the capabilities of physical servers onto a virtual system. VMWare is popular with IT departments as the platform allows companies to run many virtual machines on one server and networks can be virtualized to allow applications to function independently from hardware or to share data between computers. The storage, network and compute offered through full-scale virtual machines and Kubernetes instances for cloud-hosted applications comes with third-party support, making VMWare an unbeatable solution for enterprises.
Therefore, it makes sense Nvidia would choose VMWare’s VSphere as a partner on the Enterprise AI Suite, which is a cloud native suite that plugs into VMWare’s existing footprint to help scale AI applications and workloads. As pointed out by the write-up by IDC, many IT organizations struggle to support AI workloads as they do not scale as deep learning training and AI inferencing is very data hungry and requires more memory bandwidth than what standard infrastructures are capable of. CPUs are also not as efficient as GPUs, which have parallel processing. Although developers and data scientists can leverage the public cloud for the more performance demanding instances, there are latency issues with where the data repositories are stored (typically on-premise).
The result is that IT organizations and developers can deploy virtual machines with accelerated AI computing where previously this was only done with bare metal servers. This allows for departments to scale and pay only for workloads that are accelerated with Nvidia capitalizing on licensing and support costs. Nvidia’s AI Enterprise targets customers who are starting out with new enterprise applications or deploying more enterprise applications and require a GPU. As enterprise customers of the Enterprise AI Suite mature and require larger training workloads, it’s likely they will upgrade to the GPU-powered cloud.
Subscription licenses start at $2,000 per CPU socket for one year and it includes standard business support five days a week. The software will also be supported with a perpetual license of $3,595, but support is extra. You also have the option to have get 24x7 support with additional charges. According to IDC, companies are on track to spend a combined nearly $342 billion on AI software, hardware, and services like AI Enterprise in 2021. So, the market is huge and Nvidia is expecting a significant business.
Nvidia also announced Base Command, which is a development hub to move AI projects from prototype to production. Fleet Command is a managed edge AI software SaaS offering that allows companies to deploy AI applications from a central location with real-time processing at the edge. Companies like Everseen use these products to help retailers manage inventory and for supply chain automation.
Fiscal Q2 Earnings and More on the Arm Acquisition:
Over the past year, there have been some quarters where data center revenue exceeded gaming, while in the most recent quarter, the two segments are inching closer with gaming revenue at $3.06 billion, up 85 percent year-over-year, and data center revenue at $2.37 billion, up 35 percent year-over-year.
It was good timing for Jensen Huang to appear in a fully rendered kitchen for the GTC keynote as professional visualization segment was up 156% year-over-year and 40% quarter-over-quarter. Not surprisingly, automotive was down 1% sequentially although up 37% year-over-year.
Gross margins were 64.8% when compared to 58.8% for the same period last year, which per management “reflected the absence of certain Mellanox acquisition-related costs.” Adjusted gross margins were 66.7%, up 70 basis points, and net income increased 282% YoY to $2.4 billion or $0.94 per share compared to $0.25 for the same period last year.
Adjusted net income increased by 92% YoY to $2.6 billion or $1.04 per share compared to $0.55 for the same period last year.
The company had a record cash flow from operation of $2.7 billion and ended the quarter with cash and marketable securities of $19.7 billion and $12 billion in debt. It returned $100 million to the shareholders in the form of dividends. It also completed the announced four-for-one split of its common stock.
The company is guiding for third quarter fiscal revenue of $6.8 billion with adjusted margins of 67%. This represents growth of 44% and with the “lion’s share” of sequential growth driven by the data center.
We’ve covered the Arm acquisition extensively with in a full-length analysis you can find here on Why the Nvidia-Arm acquisition Should Be Approved. In the analysis, we point towards why we are positive on the deal, as despite Arm’s extremely valuable IP, the company makes very little revenue for powering 90% of the world’s mobile processors/smartphones (therefore, it needs to be a strategic target). We also argue that the idea of Arm being neutral in a competitive industry is idealistic, and to block innovation at its most crucial point would be counterproductive for the governments reviewing the deal. We also discuss how the Arm acquisition will help facilitate Nvidia’s move towards edge devices.
In the recent earnings call, CFO Colette Kress reiterated that the Arm deal is a positive for both the companies and its customers as Nvidia can help expand Arm’s IP into new markets like the Data Center and IoT. Specifically, the CFO stated, “We are confident in the deal and that regulators should recognize the benefits of the acquisition to Arm, its licensees, and the industry.”
The conclusion to my analysis is the same as the introduction, which is that I believe Nvidia is capable of out-performing all five FAAMG stocks and will surpass even Apple’s valuation in the next five years.