Why AMD and Intel Outran Nvidia in H1, and What Changes Now

Nvidia (NVDA) has been the default AI trade for years. The company designs the chips that power most large language models, and its data-center revenue has grown at a pace that makes other semiconductor companies look like they are standing still. Through the first half of the year, though, the stock lagged two rivals that had been written off as also-rans.

Advanced Micro Devices (AMD) rose 112% in the first six months. Intel (INTC) climbed 98%. Nvidia, by comparison, added 62%. The gap is wide enough to raise a question: did the market get the pecking order wrong, or is this a catch-up trade that is about to reverse?

The simple read is rotation. Nvidia's market cap crossed $3 trillion in June. At that size, incremental capital needs a bigger story to keep pushing. AMD and Intel started the year at valuations that assumed they would lose the AI race entirely. When both companies showed they could at least compete – AMD with its MI300X accelerator, Intel with its Gaudi 3 chip and a foundry business that is winning government contracts – the re-rating was violent.

The better read is about positioning and product cycles. Nvidia's H100 and B100 chips are still the gold standard for training. The inference market – running models after they are trained – is opening up to alternatives. AMD's MI300X has won spots at Microsoft and Oracle. Intel's Gaudi 3, due in the third quarter, is priced aggressively and targets the same inference workloads. The market is pricing in a future where Nvidia's share of AI silicon shrinks from 80% to something closer to 60% over the next two years.

That is a plausible outcome. The timing is uncertain. Nvidia's next-generation Blackwell architecture, due in the second half, could reset the performance gap. If Blackwell delivers the 2x to 3x improvement Nvidia has signaled, AMD and Intel will be chasing a moving target again. The catch-up trade works only as long as Nvidia's lead stays static.

Intel faces the hardest path. Its foundry business is a long-term bet that requires years of capital spending before it generates meaningful revenue. The Gaudi 3 chip is a product, not a platform – it lacks the software ecosystem that makes Nvidia's CUDA sticky. Intel's Alpha Score sits at 28/100, labeled Weak, reflecting the gap between its turnaround narrative and the financial reality of negative free cash flow and a dividend that was cut to zero.

AMD is in a better spot. Its Alpha Score is 49/100, labeled Mixed. The MI300X is a real product with real customers. AMD's CPU business provides a stable base that Intel's does not. AMD trades at 45x forward earnings, while Nvidia trades at 38x. The catch-up trade has already priced in a lot of success.

Nvidia's Alpha Score is 65/100, labeled Moderate. The stock is down 1.39% today at $194.83. The long-term thesis – that AI compute demand doubles every six months – has not broken. The second half will test whether AMD and Intel can deliver on the promises that drove their first-half gains. If Blackwell ships on time and the inference market grows faster than the share-shift story assumes, Nvidia could reclaim the lead. If AMD and Intel execute, the rotation has further to run.

The next catalyst is earnings season. AMD reports in late July. Intel follows in August. Nvidia reports in late August. Each print will either confirm the re-rating or expose it as premature.