The evolution of artificial intelligence (AI) has only just begun, and thus far, there have been very few companies that have been able to meaningfully tap into this nascent technology. One of the biggest beneficiaries thus far has been NVIDIA Corporation (NVDA +1.67%). The same parallel processing capabilities that enable graphics processing units (GPUs) to render images also turned out to be ideal for training AI systems.
As a result, NVIDIA is already one of the early winners of AI. The company has doubled its revenue over the last two years, partially the result of AI data center revenue that has grown 600% over the same period. Over the last 24 months, the stock has been on fire, achieving gains of 500%.
Those gains could be in jeopardy if Elon Musk has his way.
Companies are taking numerous approaches to develop the best AI chip. Image source: Getty Images.
A Tesla AI Chip in the works?
Elon Musk attended the 2017 Neural Information Processing Systems (NIPS) conference last week, and revealed that Tesla, Inc. (TSLA +2.96%) is working on custom chips for AI applications. Elon Musk recently told attendees at Neural Information Processing Systems (NIPS) conference (via The Register), "I wanted to make it clear that Tesla is serious about AI, both on the software and hardware fronts. We are developing custom AI hardware chips." Musk also reportedly said that "Jim is developing specialized AI hardware that we think will be the best in the world."
Musk was likely referring to well-known chip architect Jim Keller, who was hired by Tesla last year after stints at Apple, Inc. (AAPL +2.65%) and Advanced Micro Devices, Inc. (AMD 0.03%). Keller is credited with designing custom chips for both companies during his tenure, and as my colleague Evan Niu noted, "you don't assemble a team of rock star chip architects unless you're interested in designing custom chips for specialized purposes."
AMD, meanwhile, has been trying to up its game to compete in the field of AI, and this revelation follows reports that Tesla and AMD had been collaborating on an AI chip. While there has not been confirmation from either camp, this further supports the notion that the biggest names in technology won't sit idly by while NVIDIA reaps the spoils of its early entry into the space.
Searching for better solution
Google, a division of Alphabet, Inc. (GOOGL +1.71%) (GOOG +1.99%), and itself a pioneer in AI, revealed last year that it had designed the tensor processing unit (TPU), and that the application-specific integrated circuit (ASIC) chip had been a fixture in the company's data centers for more than a year. Google had long been a user of NVIDIA's GPUs.
Earlier this year, the search giant debuted the second-coming of the TPU, which could handle both phases of AI operations, the initial training of the system, and the execution stage, which is called inference. Previously, Google's TPU could only work on the inference phase of AI. This development could potentially ween Google off NVIDIA's GPUs.
Intel Corporation (INTC 0.07%) has been fighting the AI battle on two different fronts. Through its $16.7 billion acquisition of Altera, the company was betting that the low power consumption and flexible architecture offered by the field-programmable gate arrays (FPGA) would be useful in AI applications.
Intel also acquired AI start-up Nervana in late 2016 for an estimated $400 million. The company was working on an ASIC that removed elements of the GPU that weren't needed for AI applications. Nervana claimed it could achieve 10 times the computing power than what could be achieved by current GPUs.
It remains to be seen if either will achieve the same level of success in AI applications as the GPU.
No GPU killer...yet
The area of AI is still in its infancy and while NVIDIA gained the early advantage, there are simply no guarantees that it will be able to maintain that lead. Tesla's move to create its own hardware is merely the latest in a long line of tech icons looking to gain an edge in the race for AI supremacy.
Other companies have much to gain and NVIDIA has much to lose if someone succeeds in building a better mousetrap.
