In 2026, AI infrastructure will undoubtedly be one of the most closely watched tracks in the capital market. Everyone is chasing GPU computing power, data centers, and AI cloud services. And AI computing power rising star @ AxeCompute (NASDAQ: AGPU) is gradually filling in the most crucial pieces of the puzzle for a newly emerging Neocloud. AGPU will only transform from Predictive Oncology to an AI computing infrastructure company by the end of 2025, signing multi-year contracts worth over $3 billion in a short period of time. The growth rate is impressive, but what the market is really concerned about is whether a new player has the ability to build, deploy, and sustain such a large-scale GPU cluster. After all, signing a contract and delivering computing power are two different things. GPU procurement, data center construction, power supply, network configuration, and cluster scheduling all require funding and professional experience. For a new company, if all capabilities have to be built from scratch, expansion speed and financial pressure will become challenges. The path chosen by AGPU is different from some Neocloud players. Taking companies such as CoreWeave and Nebius as examples, they are all building their own AI infrastructure and engineering capabilities. AGPU emphasizes the use of professional platforms to complete cluster deployment and operation, focusing more on customers, contracts, and capital arrangements. This can be understood as a business philosophy of "light organization, heavy assets": not doing everything by oneself, but entrusting key links to experienced partners. The most noteworthy here is Penguin Solutions. Penguin has over 25 years of relevant experience, deploying and managing over 85000 GPUs with a cumulative GPU runtime of over 2 billion hours. It has participated in the construction and operation of Meta AI Research SuperCluster, and has also been selected as the NVIDIA cluster hosting service partner for Voltage Park. For Neocloud, which needs to rapidly expand its computing power scale, the value of such experience lies not only in building a cluster, but also in reducing the difficulty of subsequent replication and operation. From Penguin's own business performance, this route is also being validated by market demand. According to its FY2026 financial report, out of the 6 new AI infrastructure customers added in the fourth quarter, 4 were Neocloud customers. This at least indicates that more and more Neocloud enterprises are willing to rely on professional service providers to supplement their infrastructure delivery and operation capabilities. Looking at AGPU again, the current total amount of signed contracts exceeding $3 billion, customer prepayments exceeding $317 million, and a five-year 55MW custody agreement have provided support for business advancement in terms of orders, funds, and infrastructure resources. The addition of Penguin further fills in the gap of professional deployment and operation. Of course, this does not mean that AGPU is no longer risky. Prepayment does not equal profit, contract amount does not equal revenue, and the availability of data center resources does not necessarily mean that GPU clusters can be launched on schedule. Ultimately, it still depends on actual delivery, revenue recognition, and cash flow recovery. But I believe that the key to understanding AGPU is no longer just looking at how many billion dollar orders it has signed, but also whether it can convert these orders into replicable delivery capabilities through professional partners. Because the real competition in the Neocloud track is not just about building a 2000 card cluster, but whether the delivery experience of 2000 cards can be replicated to 50000 cards or even larger scales. A successful single delivery proves the ability to execute the project; Continuous replication is necessary to form the advantage of large-scale operation. Professional platforms can help new players shorten their ability building cycle, but the success of replication still depends on the cooperation of multiple aspects such as funding, supply chain, power resources, customer needs, and project management. So, the collaboration between AGPU and Penguin is not just about adding another partner, but it may have adopted a different expansion approach: AGPU is responsible for expanding customers, obtaining contracts, and organizing capital, while Penguin provides professional infrastructure construction and operational capabilities. If this model can run smoothly, the growth of AGPU will not have to rely entirely on building all engineering teams from scratch, but will have the opportunity to accelerate deployment with mature platforms. Next, whether the first batch of clusters can be launched on schedule, whether the project can start confirming revenue, and whether the delivery scale can continue to expand in the future are the real verification nodes. Orders have opened up growth opportunities, professional platforms provide execution paths, and large-scale delivery is the answer sheet that AGPU needs to deliver in the next stage.