Nvidia 2030: $625 bull case, $135 bear case

The piece frames Nvidia's future around a simple question: how much longer the AI infrastructure buildout can keep growing. The author notes that Nvidia's GPUs have a roughly five- to six-year useful lifespan, giving it a replacement cycle similar to Apple's, and argues its moat in AI model training and its position in inference should keep revenue fairly stable even if growth slows. The bear case imagines that slowdown arriving through power constraints and funding limits, plus intensifying competition from AMD and custom AI chips that could push gross margin down.

The middle case leans on analyst forecasts: revenue climbing from $411.6 billion in fiscal 2027 to $1.21 trillion in fiscal 2031, with adjusted EPS going from $9.31 to $24.79. The bull case argues analysts are underestimating Nvidia's earnings power, with demand for compute described as insatiable and cloud and neocloud companies seeing strong returns on their AI infrastructure investments. It also points to CUDA as a wide moat in AI training, the Groq acquisition as a differentiated inference play, and locked-in high-bandwidth memory (HBM) and fabrication capacity as assuring its leadership position.

The outcome hinges on whether that demand holds. If growth merely plateaus, the stock could sit near $370; if it accelerates, the bull model sees more than 2.5-fold upside; and if funding or power constraints bite, the bear case implies about 40% downside — a spread that matters most to investors weighing Nvidia at today's price.