Rising Computing Costs and Strategic Shifts in AIGC Platforms

AIGC (AI-Generated Content) is entering a new phase, shifting from being merely affordable and usable to being more refined and effective. Recently, the resale of old memberships for high prices in the second-hand market has garnered significant attention. This trend reflects the ongoing rise in computing costs and the increasing price of tokens, signaling a pivotal moment for the AIGC industry. The sector is moving away from a growth-at-all-costs model to a more value-based pricing strategy.

Price Increases: A Necessary Adjustment

As AI becomes deeply integrated into daily life and work, fluctuations in pricing are closely monitored by users. AIGC has proven its capabilities, leading to a surge in AI-generated short dramas and high-quality AI illustrations in the gaming industry. In e-commerce, AI can efficiently handle vast amounts of SKUs, accelerating the commercial viability of AI across various sectors.

However, this explosive demand is met with a critical shortage of computing power. Since the internal testing of Seedance 2.0 began around the Chinese New Year, user experiences have varied significantly, with reports of long wait times for generating content. Some AI comic companies have even adjusted work hours to submit tasks during off-peak hours in a bid to reduce wait times.

Currently, three major platforms are reporting daily token usage exceeding one trillion, highlighting the immense pressure on computing resources. Video generation models, in particular, require even more computing power, leading to inevitable delays. Since the beginning of 2026, global GPU rental costs have soared, and storage prices continue to rise, driving up model inference costs. This trend represents a hard constraint that all AIGC platforms, including Jimeng, must confront.

According to recent IDC data, China’s generative AI infrastructure market is experiencing rapid growth, projected to maintain a compound annual growth rate exceeding 60% in the coming years. This indicates a swift increase in demand for computing resources, storage capacity, and network architecture.

The recent surge in prices for Jimeng memberships is a direct result of its pricing strategy adjustments. Notably, Jimeng has ended limited-time discounts for the Seedance 2.0 model, which previously allowed members to purchase at a 40% discount during the promotional period. Additionally, they have introduced a more expensive VIP option with shorter wait times and significantly reduced monthly points for new members. This adjustment demonstrates Jimeng’s commitment to protecting existing members while implementing substantial price increases for new users.

Transitioning AIGC: From Land Grab to Refined Operations

Price increases do not equate to a loss of value. Some creators have noted that while other AI platforms may require multiple video clips to create a complete scene, Jimeng’s Seedance 2.0 can generate a seamless 15-second video, showcasing its unique capabilities.

Moreover, as video models become increasingly powerful, their cost and quality advantages become more pronounced. Compared to the tens of thousands of dollars required for live-action short dramas, the token costs associated with AIGC are minimal.

Jimeng’s price adjustment can be seen as a necessary response to rising upstream computing costs and a proactive recalibration of its business model, as well as a deeper exploration into the token economy. The token economy is gaining traction, becoming a vital component of business models. Even major domestic telecom operators like China Mobile, China Telecom, and China Unicom have emphasized strengthening token operations and expanding the ecosystem surrounding tokens.

As AI-generated content becomes integrated into commercial production processes, the criteria for measuring value are shifting. The focus is no longer solely on the cost per call but rather on whether it can yield higher production efficiency and commercial returns. B2B users, in particular, are willing to pay a premium for more stable and higher-quality services.

The choice facing AIGC platforms is clear: maintain a subsidized, low-price model that leads to resource depletion and degraded user experience, or transition to a refined operational phase where limited computing resources are prioritized for professional creators and commercial users who value stability and efficiency.

Jimeng’s adjustments represent a necessary trade-off and an attempt to address the computing pressure facing the AIGC industry. IDC emphasizes that large models are rapidly iterating and expanding, facing challenges in computing costs and data quality during both training and inference phases. As model scales grow and new technologies emerge, ensuring stability and controllability while improving training effectiveness and inference efficiency is a critical challenge for technology providers.

In the future, finding a dynamic balance between computing costs, user rights, and commercial sustainability will be a long-term challenge for all AIGC platforms.