The AI Skepticism Map: Where Short Sellers Are Betting Against AI

The financial market’s skepticism toward AI is not uniform but highly stratified, with short interest revealing a clear tension between proven AI winners and companies whose AI exposure remains speculative. Short interest across software, semiconductor, neocloud, data center, and hyperscaler stocks has risen roughly 24% in the last quarter, but the concentrated bearishness targets a specific segment: GPU data center businesses saw their shorted shares grow 60% over the past year. AI cloud and neocloud companies now carry the highest median short interest at 16.8% of float, signaling doubts about GPU value durability versus long-term returns.

Hyperscalers sit at the opposite extreme with a median short interest of just 1.1%, and NVIDIA itself is lightly shorted at 1.2%. The most heavily shorted names are small or mid-cap companies: SoundHound AI at 36.3%, C3.ai at 32.2%, BigBear.ai at 29.4%, and Applied Digital at 28.0%. Developer tools and infrastructure software at 9.5% and enterprise SaaS at 8.9% reflect a more recent negative sentiment pivot. Meanwhile, memory makers like Micron have surged 742% this year, with ecosystem CEOs pointing to memory and storage as the new bottleneck.

For builders, the key takeaway is that market skepticism is not AI fatigue—it is a bet on specificity. The largest winners are absent from the short list, while doubt clusters around companies dependent on future capital access, demand, or operating leverage. The data suggest memory has become critical and scarce, software must prove post-AI value, and GPU reselling businesses face skeptical pricing assumptions. This is not a uniform doom call on AI infrastructure, but a map of where financial markets see fragility.