When people think of artificial intelligence (AI), they think of graphics processing unit (GPU) chip leader Nvidia (NASDAQ: NVDA), and rightly so. Nvidia currently holds a significant lead in manufacturing the chips needed for today's most advanced AI applications.
But the benefits of AI will provide growth opportunities for many technology companies whose stock prices haven't skyrocketed as much as NVIDIA's.
On that note, there's big AI news this week at three big tech companies, two of which are aiming to compete head-on with Nvidia.
Arm announces it will begin manufacturing its own AI chips
Arm Holdings (NASDAQ: ARM) went public last September in a much-anticipated initial public offering (IPO), bringing to market one of its two major semiconductor architecture licensors, the other being x86. It was architecture.
Today, Arm is growing as the use cases for its low-power chip architecture proliferate. Revenue increased 47% last quarter on the back of strong growth in license and royalty income. This strong growth is not surprising. Most cloud giants have begun designing their own low-cost central processing unit (CPU) chips for their own cloud products, gravitating toward the Arm architecture. Nvidia (NASDAQ: NVDA) uses Arm-based CPUs in its Grace CPUs that connect to its Grace-Hopper and Grace-Blackwell superchips, and Alphabet (NASDAQ: GOOG) (NASDAQ: GOOGL) uses its own Introduced data center CPU chips. Axion.
But Arm hasn't gotten to the heart of the AI revolution because it hasn't licensed the GPU-accelerated architecture that's central to AI processing.
But management is trying to change that. According to a recent report from Nikkei Asia, Arm is currently considering designing its own AI accelerator to be introduced in 2025. According to the report, Arm plans to invest heavily in initial development costs, which Arm should be able to bear.It has about $3 billion in cash on its balance sheet and also has cash-generating core assets. We also carry out licensing business. According to the report, production will be outsourced to a third-party foundry, likely Taiwan Semiconductor Manufacturing.
That leaves Nvidia with more competition, which may be why Nvidia tried to acquire Arm years ago.
Still, this might be a case of Arm being a little late to the game in a crowded field. Almost all cloud giants are now designing their own accelerators. And while Nvidia maintains a strong lead in “neutral” AI GPUs, Advanced Micro Devices and Intel are each producing increasingly better “neutral” competitors.
Arm is 90% owned by SoftBank (OTC:SFTB.Y), which in turn is owned by Masayoshi Son, an ambitious technology executive. SoftBank expanded at the tail end of the 2021 tech software bubble, so Arm investors hope this doesn't become another example of “me too” thinking by entering the hot tech space too late. Must. It's no wonder they have mixed feelings about this week's announcement.
Google announces new Trillium chip
Speaking of Google's parent company Alphabet, the search and cloud giant held an input/output (I/O) developer conference earlier this week. Unsurprisingly, there were a number of AI-related announcements, including the introduction of his aforementioned Axion CPU. But perhaps most notable on the AI chip manufacturing front was the announcement of the latest version of Google's tensor processing unit (TPU) accelerator.
Google has been manufacturing TPUs since 2013, but only offers them through the Google Cloud platform. Of course, Google also has a large inventory of his Nvidia chips available for customers to use, but cloud giants are competing to find their own best low-cost alternatives for their platforms.
That's why this week's announcement of Google's 6th generation TPU accelerator Trillium was important. The new chip boasts 4.7 times more performance than the previous 5th generation chip and is also 67% more energy efficient. The new chips also feature new 3rd generation SparseCore accelerators, capable of processing twice the high-bandwidth memory with twice the chip-to-chip connection bandwidth. This will allow the new he TPU to handle larger models. Alphabet also announced the ability to string together 256 Trillium GPUs to create a Google-powered supercomputer.
The new hardware is behind a number of new AI services Google introduced at the conference, including updates to its Gemini 1.5 pro model and Large-Scale Language Models (LLM), which competes with the likes of OpenAI's ChatGPT. But Google has also introduced new models focused on specific use cases, such as his Veo for high-resolution videos, Imagen 3 for text-to-image conversion capabilities, and AI Sandbox for music and sound generation tools. did.
Finally, Google Search is also getting an AI-powered upgrade starting this Monday. Complex Google queries now show AI-generated summaries and multi-step plans for relevant searches. The new Search Overview is a great effort by Google to avoid confusion from ambitious search alternatives like Bing with ChatGPT and startups like Perplexity.
And recent results suggest Google's AI capabilities may be gaining momentum, with the company's cloud platform reaccelerating to 28.4% revenue growth in the last quarter and search becoming an aspiring disruptor. Even in comparison, it recorded an astonishing 14.4% growth on its own.
Oracle grabs big shot in the cloud
When most people think of cloud infrastructure, they think of the big three cloud platforms. But investors may not want to ignore database giant Oracle (NYSE:ORCL), which is trying to position itself as the fourth cloud alternative. Oracle's infrastructure revenue was just $1.8 billion last quarter, still far behind third-place Google's $9.6 billion. But Oracle's Infrastructure as a Service (IaaS) revenue also rose 49%, the highest growth rate among the cloud giants.
It's easy to grow from a small base, but that desire got a big boost earlier this week. Elon Musk's AI startup xAI is in talks with Oracle Cloud about a deal to rent its servers for AI processing for “several years” for up to $10 billion, according to The Information. This is a big deal considering Oracle's cloud annual utilization rate is currently $7.2 billion, making xAI one of Oracle's largest cloud customers.
While some may wonder how a mere start-up could raise such huge amounts of funding, Reuters also reports that Musk's cash has given xAI up to $3 billion in funding at an $18 billion valuation. It is also reported that it led to procurement negotiations. xAI's chatbot Grok aims to compete with Google's Gemini and his OpenAI's ChatGPT, and uses data collected from his X, formerly known as Twitter, and from Tesla cars. Visuals will have a unique angle of access to his data.
Elon Musk has had success building businesses like Tesla and SpaceX from scratch, but his acquisition of Twitter is even more questionable. Still, it looks like Oracle stands to benefit a lot if xAI turns into a large-scale LLM.
Should you invest $1,000 in Arm Holdings right now?
Before buying Arm Holdings stock, consider the following:
Motley Fool Stock Advisor's team of analysts has identified the 10 best stocks for investors to buy right now. Arm Holdings was not among them. These 10 stocks have the potential to generate impressive returns over the next few years.
Consider when Nvidia created this list on April 15, 2005… If you invested $1,000 at the time of recommendation, you would have earned $578,143. *
Stock Advisor provides investors with an easy-to-understand blueprint for success, including guidance on portfolio construction, regular updates from analysts, and two new stocks each month. TheStock Advisor service has increased S&P 500 returns by more than 4x since 2002*.
See 10 stocks »
*Stock Advisor will return as of May 13, 2024
Suzanne Frey, an Alphabet executive, is a member of the Motley Fool's board of directors. Billy Duberstein works at Alphabet and Taiwan Semiconductor Manufacturing. His clients may own shares in the companies mentioned. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Nvidia, Oracle, Taiwan Semiconductor Manufacturing, and Tesla. The Motley Fool recommends Intel and recommends options for long January 2025 $45 calls on Intel and his short May 2024 $47 calls on Intel. The Motley Fool has a disclosure policy.