American consumers are becoming more selective about where their money goes, but one type of retailer appears to be benefiting from that caution rather than suffering from it. BJ’s Wholesale Club reported a record 8.5 million members in its latest quarter, while membership-fee income increased 9.9% to $135.6 million. Digitally enabled comparable sales jumped 30%, and net sales climbed nearly 16% from a year earlier. At a time when other retailers are warning that shoppers are cutting back, BJ’s is demonstrating why the warehouse-club business model can become especially attractive when households start paying closer attention to every dollar they spend.

Normally, you might expect subscriptions and memberships to be among the first expenses consumers cancel when money becomes tight. Streaming services, software subscriptions, gym memberships and other recurring charges can quickly end up on the household chopping block. Warehouse-club memberships work differently because customers often believe the membership helps them reduce other expenses. Instead of feeling like another bill, the annual fee becomes the price of gaining access to cheaper groceries, household products, gasoline and bulk purchases. The customer is not simply asking, “Is this membership worth $60?” They are asking, “Can this membership save me more than $60 this year?”

That distinction is one of the most powerful elements of the warehouse-club model used by BJ’s, Costco and Sam’s Club. The membership creates revenue before the customer even begins shopping, but it also changes the relationship between the retailer and the customer. Once someone has paid for access, they have another reason to return because every trip helps justify the membership they already purchased. BJ’s said its growth in membership-fee income was driven by stronger member acquisition, retention and greater adoption of higher-tier memberships. The company has also historically maintained roughly a 90% renewal rate among tenured members, showing how sticky the relationship can become once customers believe they are receiving enough value.

The timing is particularly interesting because American shoppers are becoming more cautious. Recent retail results show consumers continuing to buy necessities while delaying larger purchases, shopping more selectively and spending less during individual store visits. Retailers including Walmart and Target have reported customers visiting stores while keeping tighter control over how much ends up in the basket. Consumers have not stopped spending, but they are increasingly asking whether each purchase is necessary and whether a better deal exists somewhere else.

That environment plays directly into the warehouse-club promise. A family worried about grocery prices may become more interested in buying larger quantities at lower unit prices. A commuter dealing with expensive fuel may value discounted gasoline. A household trying to stretch its budget may consolidate purchases into fewer trips or stock up on products it knows it will eventually use. The model does not require consumers to feel wealthy. In some ways, it can become more attractive when consumers feel the opposite. Economic pressure can actually strengthen the perceived reason for paying the membership fee.

BJ’s digital growth adds another layer to the model. Warehouse clubs were once built almost entirely around driving to a giant physical store and filling an oversized cart. BJ’s now reports digitally enabled comparable-sales growth of 30%, following 28% growth in the previous quarter. That suggests the traditional warehouse model is becoming more convenient without abandoning the value proposition that made it successful in the first place. Customers can increasingly combine bulk pricing and membership savings with digital ordering, pickup and delivery instead of choosing between low prices and convenience.

There is a broader lesson here for any company considering a subscription or recurring-revenue business model. The strongest subscriptions do not survive because canceling is difficult or because customers forget they are paying for them. They survive because customers believe losing the subscription would cost them more than keeping it. A business that charges $10 a month for entertainment must continually convince customers they are being entertained. A warehouse club can potentially show customers something even more measurable: how much money they believe they saved. That makes the membership feel less like consumption and more like an investment.

BJ’s record membership therefore says something larger about what makes recurring revenue durable. Consumers may cut subscriptions when those subscriptions feel optional, but they are much less likely to cancel something they believe protects their household budget. The best subscription businesses do not simply charge customers repeatedly. They create a recurring reason to stay. BJ’s latest results suggest that in an economy where consumers are becoming increasingly deliberate with their spending, helping people feel like they are saving money may be one of the most effective ways to convince them to keep spending with you.

Amazon has spent more than two decades teaching customers that delivery should keep getting faster. What once took a week became two-day shipping, then one-day delivery, same-day delivery and, increasingly, delivery within hours. Now the company wants to eliminate even more of the wait. Amazon announced this week that it plans to expand its Prime Air drone-delivery service to nearly 500 U.S. cities and towns by the end of 2026, up dramatically from just 11 operating locations today. The company says eligible packages can arrive in as little as 30 minutes, potentially turning suburban yards and driveways into the newest part of Amazon’s enormous logistics network.

The idea is fairly simple from the customer’s perspective. Shoppers in eligible areas place an order through the same Amazon app or website they already use and select drone delivery at checkout. Amazon’s newest drones can carry many products weighing five pounds or less, including groceries, electronics, cosmetics, household products and medications. Amazon says more than 60% of its most frequently purchased items meet the size and weight requirements. When the drone reaches the customer’s chosen delivery area, it checks for people, animals, vehicles and other obstacles before hovering above the ground and releasing the package. Prime members pay $2.99 for orders under $50, while qualifying orders of $50 or more are delivered by drone for free.

For Amazon, this is the latest step in a logistics strategy built around reducing the distance between inventory and customers. The company has already created massive fulfillment centers, smaller same-day facilities and increasingly sophisticated delivery networks designed to make waiting feel unnecessary. In 2025, Amazon delivered more than 8 billion items to U.S. Prime members either the same day or the next day. Drone delivery pushes that philosophy further by removing roads, traffic lights and potentially even delivery drivers from part of the final journey. A package that might normally travel from a warehouse into a van and through neighborhood streets can instead travel almost directly through the air.

But this is also where Amazon’s pursuit of convenience begins colliding with the physical world. A website can be updated almost instantly. Expanding an aviation network into hundreds of communities is much more complicated. Amazon needs regulatory and local approvals for its drone hubs, while residents and officials have raised concerns about noise, privacy and safety. Some residents have compared the sound of drones to leaf blowers, and Amazon’s rollout has already encountered opposition in certain communities. The Federal Aviation Administration provides the broader aviation framework, but each new location still requires Amazon to deal with local conditions, airspace, neighborhoods and public acceptance.

Safety becomes particularly important when thousands of autonomous aircraft could eventually be flying above neighborhoods every day. Amazon says its drones use onboard cameras and sensors to detect obstacles and make real-time flight decisions without someone remotely steering every aircraft. The company holds FAA Part 135 certification and has developed systems designed to react to unexpected weather, aircraft and objects in their path. Amazon also says its navigation cameras process information onboard rather than providing employees with a live feed of people below. Those safeguards will become increasingly important as Prime Air moves from a relatively small experiment into something Amazon hopes tens of millions of customers will eventually use.

The bigger business question is whether customers actually need everything this quickly. Amazon has repeatedly discovered that faster delivery changes behavior. Once two-day shipping became normal, five days felt slow. Once same-day delivery became widely available, waiting until tomorrow began to feel less convenient. Drone delivery could create the same effect. A customer who realizes they can receive batteries, medicine, a phone charger or an ingredient for dinner within 30 minutes may begin using Amazon for purchases that would previously have required a trip to a local store. That makes drone delivery more than a transportation experiment. It could become another way for Amazon to compete with pharmacies, convenience stores, grocery stores and other businesses whose greatest advantage has traditionally been physical proximity to the customer.

Yet the final few miles of delivery may prove harder to automate than the thousands of miles that came before them. Warehouses can be designed around robots. Software can optimize routes. Algorithms can predict demand. Neighborhoods are different. They contain children, pets, trees, power lines, changing weather, emergency aircraft, local regulations and people who may simply dislike having drones flying overhead. Amazon’s technology may work perfectly and still face resistance because logistics eventually intersects with communities rather than spreadsheets.

That is what makes Prime Air such an interesting business experiment. Amazon is no longer simply asking whether it can move a package faster. It is asking whether customers, regulators and neighborhoods are willing to redesign a small piece of everyday life around that convenience. If the expansion succeeds, seeing a delivery drone overhead could eventually become as ordinary as seeing a delivery van parked outside. If it struggles, it may demonstrate that even the world’s most sophisticated logistics company eventually encounters a boundary technology cannot easily remove. The last mile may ultimately be the hardest mile to automate.

OpenAI’s decision to slow development of an upcoming AI model because of growing cybersecurity concerns is a warning that businesses deploying artificial intelligence can no longer treat security as an afterthought.

OpenAI said it is working to stay ahead of standards for monitoring, alignment and security as AI models become more capable. The move comes as businesses increasingly use AI systems for customer service, coding, research, internal operations and increasingly autonomous tasks.

For businesses, the message is bigger than OpenAI: the more authority an AI system receives, the greater the consequences when it makes a mistake or is manipulated.

An AI agent that simply drafts an email presents relatively limited risk. An agent that can access a company’s customer database, send messages, approve transactions or modify software is different. A security failure could quickly become a financial or operational problem.

That means companies investing in AI should now budget for monitoring, access controls, testing and human oversight alongside the AI software itself.

The shift could also create a new business opportunity. As AI agents become more autonomous, companies will need tools and services that continuously test what those systems are doing and prevent dangerous actions.

The lesson for businesses is straightforward: AI adoption is moving from experimentation to infrastructure, and security has to move with it. Companies that deploy AI without building safeguards around it may discover that the cost of a failure is much higher than the cost of protecting the system in the first place.

For businesses, the next competitive advantage in AI may not simply be having the smartest model. It may be knowing how to deploy that model safely.

OpenAI is slowing the pace of development of some of its most powerful AI systems, a decision that could signal a major shift in how the AI industry approaches the next stage of the technology race.

The company said Tuesday that it is temporarily slowing certain model-development efforts while it strengthens monitoring, alignment and security. The move follows growing concerns about increasingly capable AI agents and their ability to operate autonomously in ways developers did not anticipate.

For businesses, the important story is not simply that one AI company is slowing down. It is that AI safety and reliability are becoming operational business issues rather than purely technical concerns.

Companies have spent the past two years rushing to integrate AI into customer service, software development, marketing, finance and internal operations. The next phase will require businesses to ask a harder question: what happens when an AI system is capable of taking actions rather than simply generating answers?

That changes the economics of AI adoption. A business deploying an AI agent to communicate with customers, modify records, write software or make decisions cannot treat the system like ordinary software. It needs monitoring, permissions, testing, security controls and a way for humans to intervene.

OpenAI’s decision is therefore a warning for companies building their own AI automation strategies: moving fast is no longer enough. Businesses need to build systems that can be controlled when AI becomes more capable.

The companies that benefit most from the next AI wave may not be those that deploy the largest number of agents first. They may be the ones that build reliable processes around them.

For business leaders, the lesson is straightforward: AI adoption is entering a more mature phase. The question is shifting from “What can AI do?” to “What can we safely allow AI to do on our behalf?”

That distinction could define the next era of enterprise AI.

Most consumers watch the price of regular gasoline because that is what they see every time they fill their car. Businesses may have a more important number to watch: diesel. On August 17, the U.S. diesel crack spread—the difference between the price of crude oil and the value of diesel produced from it—briefly reached a record $102.20 per barrel, crossing $100 for the first time. That does not mean diesel itself suddenly costs $102 a barrel more at the pump. It is a measure of unusually tight diesel supplies and extremely strong refinery margins, and it signals pressure in a fuel market that quietly powers a large part of the physical economy.

The surge is being driven by an unusual combination of geopolitical disruption and heavy demand. The wars involving Iran and Ukraine have reduced supplies from major refining regions. Middle Eastern fuel exports have been disrupted by refinery problems and restrictions around the Strait of Hormuz, while Ukrainian attacks have pushed Russian refining activity toward multi-decade lows and Russia has stopped international diesel sales. At the same time, agricultural demand is entering an important seasonal period as farmers use diesel-powered tractors, combines, irrigation equipment and trucks. U.S. distillate inventories have fallen to their lowest August level since 1996 even though American refineries are producing heavily and exporting fuel to markets struggling with shortages elsewhere.

This matters because diesel occupies a very different place in the economy than gasoline. Gasoline primarily moves people. Diesel moves products. Long-haul trucks carry food, clothing, electronics, building materials and almost everything else found on store shelves. Farmers use diesel to plant, harvest and transport crops. Construction companies use it in excavators, loaders and heavy equipment. Manufacturers depend on freight networks powered by diesel to receive raw materials and ship finished goods. According to the U.S. Energy Information Administration, the national average on-highway diesel price was $5.257 per gallon for the week ending August 10, more than $1.50 higher than a year earlier.

That means higher diesel prices rarely remain confined to transportation companies. Imagine a grocery product that begins on a farm, travels by truck to a processing plant, moves again to a distribution warehouse and finally reaches a supermarket. Diesel may be used at almost every stage. If the farmer’s operating costs increase, the trucking company’s fuel bill rises and the distributor pays more to move inventory, eventually someone has to absorb those expenses. Businesses may accept smaller margins temporarily, but persistent increases tend to work their way into freight surcharges, supplier prices and ultimately the amount consumers pay. USDA notes that fuel prices play a critical role across freight transportation because every mode depends on energy to move goods.

Agriculture makes the ripple effect especially easy to understand. Farmers already facing higher input costs can use thousands of gallons of diesel during planting and harvest seasons. Earlier this year, Reuters reported diesel prices above $6 per gallon in some Midwestern agricultural states as the Iran conflict disrupted energy markets. Farmers cannot simply stop using tractors because fuel becomes expensive. Instead, higher fuel expenses become another production cost attached to corn, soybeans, vegetables and other commodities before those products ever begin their journey through the rest of the food supply chain.

The current situation also demonstrates why crude-oil prices alone do not tell the entire energy story. Brent crude has fallen considerably from the highs reached during the Iran conflict, yet refined fuels such as diesel have remained expensive because the bottleneck is increasingly located at the refinery rather than the oil well. Crude oil can be available while there is still insufficient refinery capacity to turn enough of it into the specific fuels the economy needs. Global diesel exports were down sharply in July, while Middle Eastern, Russian and Chinese refining activity has all been constrained. In other words, having plenty of raw material does not solve the problem when there is not enough capacity to turn that raw material into the finished product.

For businesses, the lesson extends beyond energy markets. A company does not have to purchase diesel directly to be exposed to diesel prices. A restaurant buys ingredients delivered by trucks. A contractor buys materials transported from warehouses. A retailer receives inventory through national distribution networks. An e-commerce company depends on packages moving through fulfillment centers and delivery fleets. Even service businesses can eventually feel the impact when employees, vendors and customers face higher transportation and living expenses. Rising diesel costs are therefore less like a single expensive product and more like an additional cost gradually spreading through the economy.

Consumers may continue watching the gasoline sign on the corner because it provides an immediate picture of what driving costs them personally. But some of the most important inflationary pressure may be happening one pump over. Diesel rarely attracts the same attention because most households do not purchase much of it directly. Yet it powers many of the machines and transportation networks responsible for producing and delivering the things households buy every day. The fuel consumers rarely think about may ultimately influence the price of almost everything they do.