What Intel and Microsoft earnings say about the state of AI

Last week two specific giants in the tech industry reported quarterly earnings that could not have painted a more opposing view of the state of AI and compute. There are several more of interest coming up this week, including AMD, Qualcomm, and Amazon, but with Intel and Microsoft news now a couple of days behind us, I think its worth dissecting what those results tell us about the state of AI.

Let’s start with Intel where we found the company showing positive results with year-on-year gains in revenue, operating income and even gross margins, but showed a regression linearly. The new Intel Products group (that was just recently separated from the foundry/production business as part of a new reporting structure) revenue stream is dominated by its client division (CCG) that brought in $7.5B with $2.6B in operating income. Considering that the consumer-facing AI business and the AI PC is really just starting its upward march, I would expect the client business to continue to gain momentum with its current shipping Meteor Lake platform and the upcoming Lunar Lake designs that analysts say will ship by the end of the year.

The data center group at Intel, DCAI, continues to struggle, with $3B in revenue and $500M in operating income. This stagnation is worrisome as we have seen yet another quarter go by with consistent growth for everyone else in the world of AI compute, from Nvidia, AMD, and even smaller startup groups building accelerators. CEO Pat Gelsinger continued to promote the recent launch of Intel’s Xeon 6 family of data center CPUs and the upcoming Gaudi 3 AI accelerator solution, but even it will only attribute ~$500M of revenue to the company by the end of year. For comparison, AMD is expected to see ~$3.5B in revenue from its MI300X GPUs and Nvidia upwards of $45-50B in revenue from its AI products this year.

The fact that Q2 outlook was flat for Intel clearly showed the markets that Intel doesn’t have a clear line of sight for how it will take advantage of the AI markets for 2024 just yet, which is a bit of a let down considering its position of strength in the client space.

The story that Microsoft painted was very different, both in results and in outlook to the rest of the year. Microsoft had a total of nearly $62B in revenue and $27.6B in operating income, with a gross margin of 70%. All three of the company’s business divisions grew YoY; the Productivity group by 12%, Intelligent Cloud by 21%, and More Personal Computing by 17%.

The highlight was the growth in the Azure and Cloud Services category that saw a 31% revenue increase and alone brought in more than $26B in revenue.

The only segment that showed weakness was, interestingly, the devices group that includes the company’s Surface laptops, that saw revenue fall by 17% YoY. Much like the discussion above about Intel and its client AI PC revenue potential, the Surface group is riding on hopes that its pending system refreshes for consumers (and the recent release of new business PCs) will benefit from the AI PC phenomenon expect in the 2H of 2024.

These are interesting results, but what does this tell us about the AI market as it sits today? Almost all the massive growth and expansion of investment has come in the data center space, as evident by Nvidia’s rise to a $2T company, but also results like the ones we saw in the Microsoft earnings and its cloud division. Also, don’t forget that Microsoft is investing another $14B in infrastructure (capex) next quarter, its highest ever, so it clearly sees the revenue growth continuing.

So why isn’t Intel seeing a parallel path to its data center AI ambitions? Intel can’t continue to lean on the idea that it isn’t profitable because its investing infrastructure for foundry, as MSFT is significantly out spending them. Intel has tried pitching the market on the value of its CPUs for AI inferencing, it attempted to launch a family of Data Center GPU products to directly compete with AMD and Nvidia, and now its hoping that the next generation of Gaudi accelerators will finally make a move. Whatever product family it ends up being, the fact is that Intel is running out of runway to get customers on board with its strategy for AI.

But client and edge computing has not seen that same rise in investment or sales as the data center, at least not yet. Here we need to see tangible proof that what Intel, AMD, Qualcomm, and all their partners are saying about the AI PC and its capabilities for new computing experiences will convince consumers and businesses that they need to upgrade, to buy new machines with AI acceleration as part of the fundamental value proposition.

Microsoft Surface will be looking for that same bump from the AI PC supercycle, but it has the benefit of its data center solutions and software businesses that are clearly one of, if not THE, leaders in the clubhouse for growth in the next several years. For Intel, it needs to see success and significant design wins with Gaudi 3 AI accelerators and revenue with AI-powered client devices in the next 12-24 months to give the market confidence that its product group will flourish.