Nvidia earnings and Intel foundry prove AI still going strong

In the new world of AI, there is no more important day than today: Nvidia earnings. The market, including investors and analysts, along with competitors and partners, wait to see how well the new giant in the tech industry is faring, using it as a judgement for AI, chips, and the technology industry in its entirety.

It just so happens that today another important event in the tech field was taking place: Intel’s first ever chip foundry event in San Jose.

It should not surprise you that both of these events were focused on AI, but what might surprise is just how intertwined they are, weaving an interesting web that intertwines the fates of many of today’s tech giants.

Starting with the Nvidia earnings release, the results were excellent. It felt like the 20 minutes between market close and results release took hours, but when it finally crossed the wire, the company showed $22.1B in revenue, up 23% over the previous quarter and 265% YoY. Margins have continued to rocket upward, hitting 72%. The data center business division that is responsible for the AI products saw its revenue go from $14.5B to $18.4B QoQ; it’s even more impressive when you see a 409% increase over the same quarter last year when the data center group “only” managed to sell $3.6B of its GPUs.

For context: the largest quarterly revenue for Intel ever was just over $20.5B in Q4 of 2021.

Nvidia’s outlook for next quarter is revenue of $24B and a margin of 77%. That’s a modest revenue increase over these results, but Nvidia is probably doing its best to limit expectations and help it produce another “beat and raise” opportunity when the world meets to have this same debate 3 months from now.

But what these amazing numbers tell us is that the AI market isn’t slowing, it isn’t stopping, and people that continue to claim this is a phase or a hiccup in the market are clearly missing the significance of what real AI brings to society. Nvidia doesn’t see any slowing down of this insatiable need for AI compute, and all other indications from the tech industry corroborate.

Recently OpenAI’s Sam Altman made headlines when he said he was looking to find $7 trillion from a combination of governments and investors to rework how the world builds and distributes chips and chip manufacturing capability for AI. While most analysts believe this $7T number is more a scare tactic, meant to get the industry moving faster and investing more, I see it as a way for him prepare governments for contributing more into the ecosystem. If the federal government can get a proposal for a $100B investment in a ”CHIPS Act 2” then it could be viewed more acceptable and tame compared to the outlandish $7T number making the news previously.

Altman was one of the guest speakers at the Intel foundry event held in San Jose today. Intel brought in financial and industry analysts to showcase its continued momentum for its chip manufacturing strategy. Now officially called “Intel Foundry”, the company is now split into “foundry” and “products” divisions, with the latter responsible for building the IP for chips that power laptops, PCs, and data center servers.

CEO Pat Gelsinger stated very clearly that his intent was to make as many of the AI chips the world needs as it can. That means making chips and providing packaging for its competitors, like Nvidia, Arm, Microsoft, and many others. It’s an interesting and powerful transition: if Intel continues to struggle to find ways to compete in the AI space with its own products, where its Gaudi and GPU accelerators have had issues finding a foothold, then it can still benefit financially, and be a big player in the AI market, if it can create a leading process technology and assert itself as the preeminent domestic provider.

A couple of the more interesting partners to come on stage with Intel were Microsoft and Arm. Microsoft CEO Satya Nadella in a recorded video announced that the company had agreed to build one of its upcoming AI chips using the Intel 18A process node, though details on the volume or performance level of that chip aren’t known. But the Microsoft CEO said the entire company was committed to helping making Intel a strong player in the chip production ecosystem.

When Arm CEO Rene Haas took the stage, the language was even more concrete. Intel stated that Arm was “our most important partner” and in many ways that’s a given. A recent stat I saw mentioned that 80% of the chips that leave TSMC’s fab facilities have at least one Arm CPU core in them, so it’s imperative that Intel have the capabilities to efficiently produce Arm chips.  Arm used its stage time to highlight its recently announced Neoverse N3 and V3 cores that offer up to 50% better performance than its previous generation.

Gelsinger is placing a significant bet on its Intel Foundry business as the future of the company. It’s obvious that for every Arm Neoverse-based chip that ships that means one fewer x86 part is selling, so by building best-in-class Arm-based silicon in its fabs, it is potentially trading market share in the product space for market share in the manufacturing segment.

Both Nvidia’s earnings success and Intel’s clear dedication to being a leading foundry for AI chips are wins for other players in the AI market. AMD was up as much as 5% after hours today simply in the tailwinds of Nvidia’s continued growth, as investors (correctly) assume that the larger the market becomes, the more room for AMD products like the MI300X is created. But even the likes of Qualcomm and IBM and Google and Meta will see the global growth in AI as an opportunity. The more businesses and consumers that want AI-powered capabilities in their devices and infrastructure means more laptops and client devices with AI acceleration and more servers needed to train the AI models and feed them data.