In the AI bear case, it’s necessary to distinguish between facts and claims. Two narratives have been repeatedly circulating in the market recently. First: The argument that if memory prices fall, customers will simply forfeit their deposits and terminate long-term contracts, rendering LTA meaningless. Second: Elon Musk’s claim that SoftBank defaulted on its turbine orders. Both are sufficiently concerning issues. However, when we examine actual statements from companies, a different picture emerges. — Let’s start with memory. The past LTAs and the current SCAs Micron is signing are not the same type of contract. Micron has explicitly stated that its past LTAs were essentially non-binding supply plans. By contrast, current SCAs are different. They include multi-year volume commitments, a take-or-pay structure, where customers remain obligated to pay for contracted volumes even if they do not take delivery. Prices in many of these contracts also have defined upper and lower bounds. According to Micron’s 2026 10-K, cash deposits received from customers alone amount to $12.75 billion, with an additional $7 billion in letter of credit commitments. Thus, the notion that customers can simply forfeit a small deposit and cancel the contract if prices fall is an inadequate understanding of the current contractual structure. Of course, this does not mean memory prices cannot decline. Market prices may still fall, and profitability may weaken. Credit risk and contract disputes remain possible. But at the very least, the argument that “since contracts were broken in the past, they’ll be broken again now” must first acknowledge that the contractual structure has changed. I believe this distinction is quite important in today’s memory market debate. — The SoftBank turbine story is similar. Elon claims SoftBank defaulted on its turbine orders. Yet so far, his claim lacks specific evidence: Which turbine company? Which project? What order value? What contractual obligation was breached? None of these details have been publicly substantiated. Neither SoftBank nor any turbine manufacturer has officially confirmed such a default. More interestingly, GE Vernova’s recent earnings call addressed this directly. Morgan Stanley’s CEO asked: Are projects being delayed due to customer financing issues? Are customers requesting to cancel turbine reservations? Is the pace of contract conversions slowing? The answer was clear: No such phenomena are currently being observed. Customer financial conditions remain strong, and cash terms for securing turbines have not changed. — However, this should not be overinterpreted either. GE Vernova’s response addressed its overall customer base and backlog, not each individual SoftBank order. It did not formally disprove Elon’s claim about a specific transaction. Therefore, the most accurate characterization at this point is: The SoftBank turbine default remains an unverified claim. There is still no evidence ruling out the possibility of issues in a specific project. Ultimately, additional evidence is required. — I believe this distinction is becoming increasingly critical in today’s AI-related news. Because AI investment has become so massive, any problem—in memory supply contracts, data center financing, power supply, turbines, or credit markets— is easily interpreted by markets as evidence that overall AI CAPEX is collapsing. Indeed, risks do exist. SoftBank’s cost of credit has risen, and some data center financings are clearly under strain. But the existence of risk is not the same as actual contract failures. In Micron’s case, evidence suggests contracts are now far more binding than in the past. At GE Vernova, no widespread cancellations or financing disruptions among customers have been confirmed. Moreover, Elon Musk of Tesla is also a direct competitor of OpenAI. That does not mean his claims are false. But before accepting a single statement from a competitor as market-wide fact, it is worth double-checking actual contracts and supplier data. Currently, narratives are moving prices too quickly. Precisely because of this, it is more important than ever to focus not on who said what, but on what the actual contracts say and whether orders are genuinely being canceled.