Congratulations to AGPU for winning the big deal. After the long-term cooperation with Penguin Solutions is officially established, the market's perception of AGPU's framework should change. The core question for AGPU in the market has always been, how can a recently transformed company deliver a GPU cluster worth billions of dollars? The answer is now clear: it will be delivered by Penguin, who has built a 10000 card cluster for Meta. The order, advance payment, data center (Duos 55MW), and delivery party (Penguin) are all four pieces in a puzzle. AGPU currently lacks time, not capability. Previously, the biggest concern for institutions was always whether they could deliver or not Public materials show that AGPU has signed contracts totaling over 3 billion US dollars, and is expected to have an annualized operating rate of over 696 million US dollars after deployment. The data is very impressive. But for institutional investors, there are a series of more realistic questions here: Can these contracts be fulfilled on time? Can GPUs be purchased and delivered as planned? Can electricity, computer rooms, liquid cooling, and networks keep up? Who will build large-scale clusters? To put it simply, the market does not deny AGPU's ability to secure orders, but is concerned that its path from orders to revenue and cash flow may not be feasible. This section of the road is the most difficult to navigate on the entire chain, and it is also a source of "execution risk discount" that has always existed in valuation. In one sentence, how can a recently transformed company truly deliver a multi billion dollar GPU cluster? 2/Penguin added the most difficult piece of the puzzle Penguin Solutions (NASDAQ: PENG) is a publicly traded company on NASDAQ. It doesn't sell servers or do AI applications, it only does one thing: help others turn AI data centers from blueprints into reality, and then manage them well for the long term. What level? More than 25 years of experience in data centers and high-performance computing; Deploying and managing over 85000 GPUs, with a cumulative runtime of over 2 billion hours; Participated in Meta AI Research SuperCluster, a world-class AI cluster at the 10000 card level; It is an AI Factory Specialized Partner officially certified by NVIDIA; FY2026 revenue was 1.73 billion US dollars, net profit was 181 million US dollars (+611% year-on-year), and Q4 single quarter revenue set a record. The section added by Penguin is the heaviest, most complex, and most prone to chain failure in the AGPU chain. In short, AGPU is responsible for negotiating, purchasing, and financing the project; Penguin is responsible for building, running, and ensuring long-term success of the project. This means that AGPU does not have to start from scratch with a large engineering team, but instead relies on a construction team and operations team that has been verified countless times to ensure a stable delivery and operation. The most concerning question for institutions is' who will build it and whether it can be built ', and now there is a clear answer. 3/The four pieces of the puzzle are complete, discuss whether the center of gravity should be shifted backwards Now, we have the order, the advance payment, the data center, and the delivery party. The four pieces of the puzzle are complete. The focus of the discussion will naturally shift to the next stage: Can the first batch of projects be constructed and launched on schedule? Can we start generating verifiable revenue and operational data? This is a typical transformation of AI infrastructure projects from storytelling to homework submission. The story stage is compared to imagination, while the execution stage is compared to fulfillment.