How will the AI giants actually perform? The Street identifies nine companies as “giants” in the space, and of course they have their own view on who among these is most likely to tower above the rest. So do enterprises, and so do I. Let’s look at the companies and see if we can build a position with logic.
What we’re going to do for each company below is to assess their AI exposure, what specific AI developments they depend on for long-term success, what they’re doing to firm these developments up in the market, and how likely that is to be successful. We’ll take the companies alphabetically.
You’ll note that the so-called “magnificent seven” are all on the list. They are Nvidia (NVDA), Microsoft (MSFT), Alphabet, Amazon (AMZN), Meta Platforms (META), Apple (AAPL), and Tesla (TSLA). The Seven are poster-children for the Street’s AI angst; the group has under-performed significantly, with a P/E valuation premium relative to the S&P 500 at its lowest level in over 10 years. Only Google outperformed the S&P 500 this year, and you’ll see why below.
Amazon is only indirectly exposed to AI, for the obvious reason that it has a wide range of businesses. Its AI push is through AWS, where it’s promised a $200 billion AI infrastructure investment. The company’s AI focus, then, is really the enterprise, and its cloud business is an obvious path to business AI services. Amazon also understands that enterprises have not been willing to surrender core business data to the cloud for governance reasons, but it’s not yet clear that they have an answer to those concerns. One possible strategy is to focus on government as a vertical; they already have a strong Federal systems approach, and they’re rumored to be looking more at state/local and international government customers. Wall Street is still concerned about their ability to monetize their AI investment fully.
Anthropic is one of two AI giants that is not yet publicly traded, though it has filed confidentially for an IPO later this year. It’s focus, with its Claude model, is enterprise agentic knowledge worker support. Its tools tend to act as autonomous assistants to key workers, which enterprises tell me would make up somewhere between 8% and 15% of their workforce, depending on the vertical and the nature of the business. The big challenge they face is governance, since this mission could easily involve using company-private data. Like OpenAI, they’ve been making announcements to increase the buzz around AI, including a recent one on “J-space” an internal mechanism they claim to have found in their model, which might explain how AI works.
Apple may be one of the two smartest players in the AI game. They recognize two things. First, we still can’t identify roles that users or workers would pay for, to generate a return on AI investment. Second, it’s always a risk to ignore a hype wave. Their response has been to respond to AI publicity with relationships (like the recent deal with Google) that minimize their direct investment but still satisfy the demand for them to demonstrate an AI strategy. Historically, Apple is more focused on the consumer/individual than on the enterprise, which could be a risk for them because consumer willingness to pay for AI is likely harder to develop.
Google is, IMHO, the smartest player in the AI game. Android and Pixel give them a major tool in promoting personal and “edge” AI. Their search business gives them a consumer position, and their cloud business an enterprise position. Google’s workspace tools, from documents to email, are an on-ramp to focused AI agents. Their Gemini models are among the best, by enterprise rating, and their agent focus on documents, videos, audios, and images mean that they can apply AI to the creation of personal and business material fairly easily. Their cloud business gives them a strong on-ramp to evolving IoT and world-model missions, too. So far, they have not focused on wearables, which means that could be a risk if those devices play a major role.
Meta has perhaps the biggest challenge of the giants. Their real strength lies in Facebook, a consumer offering, and in their “metaverse” direction. The problem is that consumer AI is the most problematic source of ROI, and while the metaverse is perhaps the first full articulation of a world model, the company failed to develop the initiative properly when it was announced, and so a difficult re-launch would be needed to make it more broadly useful, especially to enterprises. Meta has AR/VR glasses, though, and these are arguably a major piece of any real-world AI application, even for enterprises.
Microsoft is largely focused on business use of AI, and like others with this goal, is focusing more on agents designed to be embedded in workflows. That’s consistent with how enterprises see their own goals, but whether than can succeed is again highly dependent on governance/trust resolution. The company has historically linked to OpenAI models, but has been self-developing smaller models that could, in theory, be distributed to the user level. Microsoft Azure is perhaps the most credible enterprise cloud service, and Microsoft’s Office and Teams tools give it a direct link to workers, but is it at risk to data governance? Is it trying to use AI hype to boost its cloud business? Maybe, but they have a good enterprise shot.
Nvidia is in an interesting position. It’s current revenue is almost totally dependent on cloud AI giants, but that makes it dependent on the cloud model and almost an automatic enemy of self-hosted AI. At the same time, though, they’ve done more work on world models, digital twins, IoT applications, and even enterprise self-hosting than most of the other giants. I think they know that the current hype wave will crest and fall, but they also know that any attempt to promote a successor concept like self-hosting will hasten the dip and likely hit their bottom line, temporarily. Thus, they’re likely to continue to promote the hype while preparing for it to fail.
OpenAI seems to have a dual strategy for AI. The “real” track is toward enterprise AI agents, cloud-hosted and providing the same sort of knowledge worker support that Anthropic targets, buttressed by the creating of custom models and integrating with partner companies to create an ecosystem. The “hype” track is aimed at keeping the company’s name in the media by talking consistently about things like artificial general intelligence (AGI). OpenAI is prepping for an IPO, but they recently developed a major problem that could impact their IPO and credibility overall, a lawsuit from Apple alleging theft of trade secrets.
xAI (including Tesla, SpaceX, and so forth) seems to be focusing more directly at the consumer, despite the fact that some of the elements of Musk’s AI empire are clearly non-consumer (SpaceX). All of the technologies in the empire support each other; Tesla and SpaceX gather or will gather data that populates the AI tools for training. He is also working to create “virtual employees”, which take personal applications into worker applications. Musk’s own mindset is the major driver of policy and innovation here, which is potentially an advantage and also a risk. His breadth of products relating to or using AI means he can likely take the ecosystem in many different directions depending on market requirements.
Which of our giants is truly gigantic? If that was a question with an easy, solid, answer, everyone would be hunkered down on it. The fact is that there are three ways AI success could be achieved. First, sell it to consumers. Second, sell it to workers (perhaps through their business) directed at personal productivity, and third, sell it to businesses to somehow improve operations at a higher level. Right now, the focus has been to pluck the low apples, which means go after the consumers and workers who have the greatest willingness to pay, the highest economic value. Are there enough of these to make AI a big long-term success? I’ve never believed there were enough, and I still don’t. The giant of the future has to leverage another path, one we can’t yet prove out.
My view is that Google, Apple, and Microsoft have the clearest path to AI success, simply because they seem best aligned to weather the hype-to-reality shift that’s surely coming, meaning they can dance in at least one credible direction. Meta and xAI have a big upside and almost equally big downside depending not only on market direction but on their own actions, which are difficult to predict. Nvidia’s success depends on the delicate balance between exploiting the current hype and dealing with its inevitable crash without actually hastening that event. For the rest, it’s simply too soon to tell how they’ll shake out.
