Recent news coming out of Meta Platforms (META 2.52%) and Amazon (AMZN 1.52%) offers the latest proof for why Alphabet (GOOGL 1.66%) (GOOG 1.68%) is one of the best artificial intelligence (AI) stocks to own.
According to a report from The Information, Meta canceled the development of its most advanced custom AI chips after struggling with its design. It had planned to use the chips to help train its AI models. Instead, the social media giant will focus on a simpler version. This is the second AI training chip design that Meta has scrapped, demonstrating how difficult it can be to design these chips, even with the help of a company like Broadcom.
Image source: Getty Images.
Meanwhile, Amazon exec Peter DeSantis revealed to The Wall Street Journal that the cloud computing leader was planning to develop its own AI models trained on its custom chips. DeSantis noted that if the company can successfully build AI models using its own custom chips, it would be able to do so at a fraction of the cost of a pure-play large language model (LLM) provider like OpenAI and Anthropic. However, after a $50 billion investment in OpenAI's latest funding round and an already large investment in Anthropic, the company is also clearly hedging its bets.
NASDAQ: GOOGL
Key Data Points
Alphabet is the AI company to beat
However, this news shows why Alphabet is currently the AI stock to beat. Creating custom AI chips isn't a piece of cake, as Meta has shown, and Alphabet has a large lead in the space. Its highly regarded Tensor Processing Units (TPUs) were first developed over a decade ago and are now on their seventh generation. They run most of the company's internal workloads and thus have been battle-tested. Meanwhile, the chips are so good that several big AI companies, including Meta, Anthropic, and Apple, are renting the TPUs to help run their own AI applications.
Meanwhile, Amazon's comments about looking to use its in-house chips to develop an AI model demonstrate the current advantage Alphabet has over its AI model competitors. By using its TPUs to train its Gemini model, it just has a huge structural cost advantage over competitors like OpenAI and Anthropic. Importantly, though, this is just going to grow as more computing power is thrown at these AI models. Alphabet plans to spend a massive $175 billion to $185 billion on capital expenditures (capex) related to AI data centers this year, but what is often overlooked is that it is getting a lot more bang for its buck with its spending than companies relying on Nvidia's graphics processing units (GPUs).
With the most complete tech stock in the game, Alphabet is the AI leader and a stock to own for the long term.
