The recent run of AI news points to a single pattern: the technology's next phase is gated less by capital and compute than by permission. SoftBank is seeking $100 billion from Middle Eastern investors for AI-refined projects, as Tom's Hardware reported, while Amazon has dropped nondisclosure agreements in data center negotiations and startups are betting consumers will hand AI agents access to their credit cards, per TechCrunch. Money and models are abundant; trust, access, and local approval are not.
The NDA Retreat Is a Trust Problem
Amazon's decision to stop using NDAs when negotiating data center deals with local governments, following a similar move from Microsoft earlier this year, is best understood as an admission that secrecy has become a liability. As TechCrunch reported, that secrecy has fueled community backlash against AI infrastructure, with opposition leading to hundreds of proposed and enacted moratoriums from New York to San Francisco.
The numbers matter less than the mechanism. A moratorium is not a market verdict; it is a procedural one. It says that the people who will live beside a data center do not believe they have been given enough information to consent. Nondisclosure agreements, whatever their commercial logic, read to those communities as evidence that something is being hidden. By dropping them, Amazon and Microsoft are not conceding that data centers are harmful. They are conceding that the negotiation itself had become the bottleneck.
For US technology companies, this is a strategic shift. The constraint on AI capacity is no longer only chip supply or grid interconnection. It is the willingness of county boards, city councils, and state legislatures to say yes. That willingness is now a function of disclosure, and disclosure is a function of how much companies are prepared to reveal about water use, power demand, tax treatment, and local employment. The moratoriums from New York to San Francisco are the price of having withheld that information.
SoftBank's Bet Rests on the Same Assumption
SoftBank's plan, reported by Tom's Hardware, is to buy companies that do not use AI or robotics and then apply those technologies to improve their operations and valuation. It is seeking $100 billion for the effort from Middle Eastern investors.
The strategy is a bet that AI's returns will come not from new products but from retrofitting existing businesses. That is a permission problem in a different register. A retailer, a logistics firm, or a manufacturer that is acquired and then automated must persuade its workforce, its suppliers, and often its regulators that the change is legitimate. The capital is available. The consent is not guaranteed.
It also concentrates risk. If SoftBank is buying companies explicitly because they lack AI and robotics, it is buying businesses whose value depends on the labor and processes it intends to replace. The improvement in valuation depends on the replacement succeeding. That is not a technical question; it is an operational and political one.
Agents Need a Different Kind of Yes
Meanwhile, a wave of startups is betting that consumers will hand AI agents access to their credit cards, as TechCrunch reported. This is the most direct permission question of the three. A data center moratorium is a collective decision made by a local government. An agent with a credit card is an individual decision made by a consumer, repeatedly, under uncertainty.
US consumers have spent years being told that their financial data is valuable and that they should be careful with it. Asking them to delegate purchasing authority to a software agent requires a level of trust that the industry has not yet built. The startups in question are not wrong that the convenience could be substantial. They are wrong if they assume that convenience alone will overcome the instinct to keep spending under direct control.
