AI Briefs
The soaring AI bills are reshaping enterprise model selection strategies.
As AI usage costs skyrocket, enterprises shift from pursuing the most powerful models to prioritizing cost-effectiveness, ushering in opportunities for open-source and domestic models.
Industry Context
In the past few years, enterprises have rushed to deploy the most advanced AI models to stay competitive, but soaring bills are now forcing them to rethink. According to iTnews, Uber's 2026 AI budget was used up by employees within four months due to heavy usage of AI coding tools, prompting management to restrict usage. Meanwhile, AI vendors are shifting their pricing models from fixed subscriptions to per-token consumption, making costs unpredictable for businesses. BlueRock CEO Harold Byun revealed that many clients report budget overruns of 20%–30%, stating, "The change in pricing models has caught many off guard."
Market Impact
Soaring costs are straining corporate IT budgets. Gartner predicts that by 2028, AI coding costs will exceed average developer salaries; its survey also shows that 75% of executives expect technology budgets to rise this year, with nearly half forecasting double-digit increases. Companies are beginning to embrace cheaper models: data from OpenRouter (an AI model marketplace) shows that the share of open-source tokens jumped from 34% in January to 65% in June. Enterprises are increasingly adopting routing tools to assign tasks to the most cost-effective models, reserving complex work for high-end models.
Competitive Landscape
Increased cost sensitivity is directly reshaping the competitive landscape. The beneficiaries are open-source model providers, particularly China's DeepSeek. According to a Citi report, Chinese models like DeepSeek occupy the top four spots on OpenRouter, with prices just 1/22 of top-tier US models — 18 cents vs. $4 per million tokens. Palo Alto Networks CEO Nikesh Arora has publicly urged AI labs to "price in advance," charging expected future prices at current lower rates. OpenAI is reportedly considering significant price cuts, including lowering token usage fees, to compete with Anthropic. However, price reductions could harm these companies' revenue growth, especially as they prepare for potential IPOs.
Enterprise Implications
Enterprises should develop cost awareness and introduce AI routing tools for tiered task management. For sensitive industries, security concerns around Chinese models remain a barrier, but open-source models offering "90% of the performance at 10% of the price" are attracting adoption for non-core use cases. It is recommended that companies learn from cloud computing's multi-cloud strategy, diversifying risk across multiple model providers and flexibly selecting based on task complexity.
OutlookIn the next 12-24 months, the capability gap of open-source models will narrow to about 4 months, and price wars may force OpenAI, Anthropic, etc. to accelerate IPOs to obtain funding, but valuations may come under pressure due to profit compression. Looking further ahead (3 years), the AI market will move towards a multi-cloud, model-commoditized ecosystem, and enterprise AI procurement will shift from "choosing the strongest" to "choosing the most cost-effective."
Article context · aiindustryreview
aiindustryreview frames this note through AI Models / Model releases and capability claims / Evaluation, safety, and benchmark signals. AI Models / Model releases and capability claims / Evaluation, safety, and benchmark signals explains the local editorial angle; dates, names and status changes still need checking. Source links should be opened before the summary is reused.