Post

#AIInfraEarningsWatch AI infrastructure is entering the part of the cycle where revenue growth alone isn't enough anymore. Investors now need to see whether enormous AI spending is actually turning into sustainable earnings. The numbers are still impressive. NVIDIA's latest reported quarter produced $81.6B in revenue, up 85% YoY, while Data Center revenue reached $75.2B, up 92% YoY. AMD is also showing how quickly demand is moving toward the data center. Its 2025 Data Center revenue reached $16.6B, up 32%, driven by EPYC CPUs and Instinct accelerators. But here's where I think the market is becoming more selective. AI infrastructure isn't just GPUs anymore. The stack now includes: → GPUs and accelerators → CPUs → networking → optical connectivity → memory → cooling → power generation → data-center construction And the bottleneck is increasingly moving beyond the chip. That's why upcoming earnings deserve attention beyond headline EPS. I'm watching three things: 1. Revenue conversion Are AI infrastructure orders becoming actual recognized revenue? 2. Capex efficiency How much capital is required to generate each additional dollar of AI revenue? 3. Customer concentration If a few hyperscalers are responsible for most demand, what happens if their capex growth slows? My opinion: The AI infrastructure thesis is still very strong, but the easy phase may be ending. When almost everyone agrees that AI spending will grow, the next question becomes: Who actually earns an attractive return on all that spending? That's where the next phase of the AI trade gets interesting. #OKXOrbitTopics #OKXTraderVoices $XNVDA $NVDA $AMD