The AI era has arrived; from now on, the core competitiveness is no longer manufacturing, but crushing the competition from a higher dimension!
The AGI era begins: the core focus of the fourth industrial revolution
On September 3, 2026, OpenAI officially released its flagship model, GPT-6 Astra. President Greg Brockman stated bluntly at the launch event: "Welcome to the AGI era." This is not marketing talk—Astra achieved 97.6% on the FrontierMath Tier 4 math benchmark and 99.9% on ARC-AGI-3 general reasoning, and it also has the ability to autonomously operate computers to complete end-to-end tasks. The model was trained in Texas on the "Stargate" supercomputer using over 100,000 GPUs. Its API pricing is 10 for input and 50 for output per million tokens, and it is the first model whose training was supervised with the participation of previous AI models.
From the steam engine to electricity, from the internet to AI—the Fourth Industrial Revolution has officially entered a substantive phase. AGI is not a concept; it’s the engine that is actively reshaping productivity. The following eight targets are the most direct beneficiaries of this transformation.
Compute layer: NVIDIA / Super Micro Semiconductor (AMD)
AI’s “oil” is compute, and GPUs are the refineries. NVIDIA remains the undisputed leader; its Blackwell architecture and subsequent products continue to define industry standards. AMD is also racing hard with its MI-series accelerators, grabbing share in inference-cost-sensitive scenarios. Together, the two form the compute foundation for AGI.
Storage layer: Micron / SanDisk
Astra’s 1.05 million-token context window and multimodal training are driving exponential growth in demand for high-bandwidth memory (HBM) and high-speed SSDs. Micron is a core supplier of NVIDIA HBM; its HBM3E and next-generation products directly determine the production capacity ceiling of AI servers. SanDisk has been spun off from Western Digital into a pure NAND flash play; by 2026 its stock gain exceeds 720%. Its 256TB enterprise SSDs are being adopted at massive scale by hyperscalers to address storage bottlenecks in AI data lakes.
Energy layer: Bloom Energy
A 100,000-yuan-worth GPU computing power cluster: the biggest constraint is not the chips, but electricity. Bloom Energy’s solid oxide fuel cells are becoming an “on-site” power station for AI data centers, helping operators bypass grid capacity bottlenecks. Its share price has surged 441% over the past 52 weeks, as data center operators view it as a bridging energy solution for AI infrastructure.
Network layer: Lumen Technologies
Compute needs connectivity. Lumen Technologies has transformed from a traditional telecommunications provider into the “fiber backbone” of the AI economy. Its Private Connectivity Fabric (PCF) contract total is nearly $13 billion, providing dedicated fiber connections between AI data centers for hyperscalers such as Anthropic and Microsoft. CEO Kate Johnson puts it plainly: “Our expansion isn’t just fiber miles—it’s about achieving the fastest time-to-first-token response.”
Hardware & application layer: Dell / Tesla
Dell is a direct supplier of AI servers, with enterprise GPU server orders continuing to ramp up. Tesla is the ultimate scenario for AGI deployment—from FSD autonomous driving to the Optimus humanoid robot. Its Dojo supercomputer and end-to-end neural networks are, in themselves, the largest physical-world testing ground for AGI.
The launch of GPT-6 Astra is a watershed moment. When models can operate computers autonomously, discover zero-day vulnerabilities, and solve cutting-edge mathematical problems, “intelligence” itself has become a tradable commodity. The eight targets above cover every link in the AGI industry chain—compute, storage, energy, network, hardware, and applications. The investment logic of the Fourth Industrial Revolution has never been clearer: it’s not a bet on who will win the AGI race, but a bet on the underlying infrastructure that enables AGI to run.