Global memory production capacity is being swallowed up by AI data centers, and automated scalping systems have begun to "scratch ground" along the supply chain. From retail DDR5 strips to industrial-grade memory modules and even connector components, they have become targets of monitoring and arbitrage.
A recent operation disclosed by DataDome, which is responsible for robot risk control, shows that a set of DDR5 scalper crawlers can quietly refresh product pages every 6.5 seconds without triggering conventional risk control thresholds, initiate an average of more than 550 automatic visits to a single page, and accumulate more than 50,000 requests per hour on multiple target sites. DataDome says it has intercepted more than 10 million access requests from this single bot, all directed to the latest price and inventory data for DDR5 memory and related components.
This set of crawlers targets far more than the finished memory SKUs that consumers can see. DataDome's analysis indicates that the bot enumerates multiple levels of the memory supply chain: from standard DIMM slots and CAMM2 press-fit memory connectors to industrial-grade memory modules typically sold in B2B environments. This behavior means that scalpers are not satisfied with reselling retail sets, but also hope to monitor the supply constraints of upstream parts in real time and obtain leading signals before supply is tight.
In order to reduce the risk of being identified, the operator deliberately simulated the "circadian rhythm" traffic pattern: the request volume showed a daily cycle over time similar to human Internet habits, but the overall curve was extremely smooth and finely calibrated, and the access frequency was always below the volume alarm threshold adopted by many e-commerce and distribution platforms. Each HTTP request will be accompanied by unique parameters to "bypass the cache", forcing the server to treat it as a new visit instead of returning the cached page, thus ensuring that the latest prices and inventory are obtained instead of lagging data. These sessions are extremely short and robotic—visiting a single product page and leaving immediately, with no adding to cart, site search, or lateral browsing behavior.
Although the pace is close to human browsing, this set of traffic still exposes obvious automation characteristics. Visits are almost entirely concentrated on memory products and will not spontaneously spread to other hardware categories on the site. The traffic curve also lacks the "weekend trough" or "evening peak" of normal users; on the contrary, when the robot encounters a technical failure, the traffic will instantly drop off a cliff, and then immediately return to full capacity after repair. This "step" change is very different from the gradual response of human users to failures and delays.
The action’s focus on DIMM slots highlights the sensitivity of the DDR5 ecosystem at the component level. Traditional desktops and servers still rely on standard DIMM slots. Therefore, detecting the supply of basic components such as slots can determine in advance which platforms and manufacturers will face tighter supporting supplies. The new generation CAMM2 standard is also within the scope of monitoring. It provides high-density memory configuration in a thinner form, mainly for notebooks and compact systems, and gradually enters some edge computing scenarios. Continuous monitoring of CAMM2 connector and module inventories provides scalpers with a view into a market segment with a relatively narrow supply and higher design concentration.
Industrial-grade memory modules are included in the crawling target from another dimension. Such products are aimed at embedded, network equipment and industrial PCs, emphasizing wide temperature range, long-term supply and high reliability, rather than extreme bandwidth. The crawling of industrial SKUs and B2B directories means that scalpers do not limit their vision to game consoles or enthusiast platforms, but are simultaneously tracking the underlying hardware that supports infrastructure and operational technology.
This round of action in the DDR5 field continues the scalper model that has been common in popular hardware and limited products in recent years. Scalpers have marked up the price of Sony's PlayStation 5 Pro 30th Anniversary Edition pre-sale, with resale offers on the secondary market ranging from three to six times the original price. High-end GPUs have also followed a similar trajectory: within days of the NVIDIA RTX 5090 being launched, some cards were priced at two to three times the official suggested retail price on the secondary market, while the limited edition MSI RTX 5090 Lightning Z was even listed for a premium of nearly 500%.
DDR5 itself has long been a breeding ground for aggressive arbitrage. An analysis of eBay listings for high-end strips shows that some products have retail prices soaring from about $118 to about $430 before being resold by scalpers on the secondary market for more than $830, or more than seven times the original price before supplies were tight. Some other high-spec configurations have been sold for well over $2,000 after multiple rounds of price increases.
Underpinning all of this is the structural tilt of the memory supply pattern toward hyperscale data centers and AI workloads. Industry forecasts indicate that data centers will consume about 70% of global memory chip production by 2026, a shift that has substantially squeezed inventory in other market segments, ranging from consumer PCs to automotive electronics to consumer electronics products that rely on a mix of old and new DRAM. Manufacturers such as Micron, Samsung, and SK Hynix are under pressure to accelerate the construction of new factories and expand production lines. However, new memory factories can often be completed in years, not months, from project establishment, equipment installation, to ramp-up mass production. During this period, any gaps at the wafer level will be quickly amplified along the supply chain and transmitted to slots, connectors and final finished modules - precisely the nodes that such scalpers closely monitor.