Microsoft (MSFT +2.20%) shares are down around 20% to start the year, as of this writing. The company was caught up in worries about the impact of artificial intelligence (AI) on software-as-a-service (SaaS) companies and over its AI infrastructure spending.
While Microsoft's enterprise software business should largely be protected from AI disruption, given how ingrained it is, and its cloud computing is growing fast and has big commitments from OpenAI, it's not the best opportunity in the cloud space, in my view.
The reason is that the company has neither a top AI model nor top-tier custom AI chips, which leaves it at a disadvantage. Microsoft was smart to invest and team up with OpenAI at the start of the AI boom, but its reliance on its models has also set it back a bit. Meanwhile, it is still very early in the process of developing custom AI chips, and remains behind in this area, as well.
Instead, the better opportunity in the market right now is with Alphabet (GOOGL 0.31%) (GOOG 0.51%) as well as Amazon (AMZN +0.30%). Here's why.
Image source: Getty Images.
1. Alphabet: The complete AI package
While only the third-largest cloud computing provider, Alphabet is arguably the best positioned moving forward. The company's biggest advantage is that it has developed the best custom AI chips, called tensor processing units (TPUs), which give it a huge cost advantage.
With the help of Broadcom, Alphabet developed its TPUs more than a decade ago and has optimized its entire hardware and software stack around these chips. Now in their seventh generation, its TPUs are battle-tested, given that Alphabet has run most of its internal workflows and trained its highly regarded Gemini foundational large language model (LLM) using these chips.
In addition to the cost savings the company gets from AI model training and running inference with its own chips, Alphabet is also set to benefit from customers deploying its chips, both within Google Cloud and in their own data centers. For example, Anthropic has already placed a $21 billion TPU order through Broadcom to be delivered this year, and the trio recently announced an expanded deal starting in 2027.
NASDAQ: GOOGL
Key Data Points
At the same time, Alphabet has embedded its TPU-trained model, Gemini, throughout its products, including Google search. This has made them better and is driving query and revenue growth. With a powerful ad network and a big distribution advantage, Alphabet also has the ecosystem in place to really drive growth in its core business with the use of AI.
With the most complete AI stack, Alphabet is a great AI stock to own for the long haul.
2. Amazon: The cloud computing leader
Amazon is the largest cloud provider, having created the entire infrastructure-as-a-service concept with Amazon Web Services (AWS). Like Alphabet, it also has a large custom chip business. While its chips are not on the same level as TPUs or Nvidia's graphics processing units (GPUs), they are solid bargain chips that are starting to gain traction. The company recently said that this was a $20 billion run-rate business growing triple digits, and noted that it would be a $50 billion business if it included the chips it deploys for internal use.
NASDAQ: AMZN
Key Data Points
The company said its Trainium chips will save it tens of billions of dollars a year in capital expenditures (capex) and boost its operating margins given their cheaper cost to run inference. Meanwhile, Amazon has its own custom central processing units (CPUs), which should be a big advantage, as CPUs are set to be the next big AI infrastructure bottleneck with agentic AI. This all positions AWS very well for the future.
Meanwhile, Amazon's e-commerce business continues to hum along. Its investments in AI and robotics are driving huge efficiencies that are leading to strong operating leverage. With the dual growth driver of cloud computing and e-commerce, Amazon is a stock to own for the long term.
