Businesses Turn to Open-Weight AI Models to Curb Soaring Tech Costs

Enterprises are increasingly evaluating open-weight artificial intelligence models to balance computing capabilities with operational expenses.

TL;DR

Running open-weight models on proprietary hardware allows companies to avoid per-query API fees, though it requires managing local infrastructure.

• Surging Expenses: Rising costs for proprietary frontier AI services are prompting businesses to adopt lower-cost open-weight models.
• Sixfold Increase: Mentions of open-weight and open-source models in corporate earnings calls jumped sixfold in August and September compared to the same period in 2025, according to AlphaSense data.
• Cross-Industry Adoption: Companies such as Tinder, Siemens, PNC Financial Services, and C.H. Robinson are exploring or deploying open-weight models to control budgets.
• Infrastructure Trade-Off: While open models cut recurring vendor fees, businesses must weigh these savings against the cost and operational responsibility of managing their own computing infrastructure.

Escalating AI Expenses Drive Corporate Shift

The choice between open-weight and proprietary models is becoming a strategic financial calculation for chief financial officers and technology executives.

American enterprises are increasingly adopting open-weight artificial intelligence models as corporate leaders seek to rein in soaring technology expenses. According to a report by the Financial Times (FT), companies are finding that running open models on their own hardware can offer a substantial cost advantage over paying ongoing access fees for closed, proprietary systems developed by providers like OpenAI and Anthropic.

Corporate leaders are closely monitoring AI operational costs as enterprise spending on frontier models surges.

The financial pressure is accelerating a noticeable shift in executive strategy across multiple industries. Data from research platform AlphaSense reveals that references to “open weight” or “open source” models during corporate earnings calls and investor conferences jumped sixfold during August and September compared to the same two-month period in 2025.

Broad Cross-Sector Interest Beyond Tech

While software and technology firms were among the earliest to experiment with alternative model architectures, interest in open-weight AI has quickly expanded across traditional industries. Recent executive discussions regarding open-weight adoption have emerged from banking institution PNC Financial Services, logistics company C.H. Robinson, and industrial giant Siemens, according to the FT report.

Case Study: Tinder Tackles a Tenfold Cost Spike

The economic reality facing enterprises was highlighted by Vinay Kuruvila, chief technology officer at dating app Tinder. Kuruvila reported that the company's annual AI expenditure rate surged dramatically from $1 million in January to $10 million by July, prompting leadership to take proactive measures to avoid another tenfold jump in costs.

To control these escalating expenses, Tinder began routing selected queries from non-technical users to open-weight models rather than relying exclusively on commercial frontier offerings. Discussing the transition, Kuruvila noted:

“In January we were spending at the rate of $1 million per year and by July it had climbed to $10 million … I don’t want another 10X increase. The frontier models like OpenAI’s Astra and Claude Fable already have enough intelligence for 90% of the tasks we’re trying to do. If open-weights models catch up, I may not need to use them anymore.”

Understanding Open-Weight vs. Proprietary AI Models

To understand why businesses are reconsidering their AI procurement, it helps to distinguish how different models are delivered and monetized:

• Proprietary Frontier Models: Providers such as OpenAI and Anthropic offer access to high-capability models (such as OpenAI's Astra or Anthropic's Claude Fable) through cloud-hosted APIs. Companies pay per token or query, which can lead to rapidly escalating operational costs as user volume expands.
• Open-Weight Models: In an open-weight framework, model weights are available to users, enabling businesses to download and run the models directly on their own data center servers or private cloud hardware. This eliminates per-query vendor charges, though it requires internal computing infrastructure.

The Strategic Infrastructure Trade-Off for CFOs

As PYMNTS reported, the growing interest in open-weight models represents a financial calculation rather than an ideological debate for corporate financial officers. While earlier debates centered on open-source philosophy, middle-market chief financial officers (CFOs) are primarily evaluating cost efficiency, operational autonomy, and infrastructure demands.

According to PYMNTS analysis, the decisive issue is not whether open-weight models will eclipse proprietary AI, but whether the cost savings, flexibility, and control they deliver are sufficient to justify taking on direct responsibility for hosting and maintaining underlying computing hardware. This calculation is expected to become even more vital as enterprise AI moves beyond isolated chatbots into core business functions such as finance, procurement, treasury, compliance, and enterprise software.

Open Infrastructure and AI Defense Collaboration

The broader debate over open architecture in artificial intelligence has also attracted attention from major technology suppliers. Earlier this year, Nvidia partnered with several tech leaders to establish a new AI safety coalition. In its official announcement, Nvidia called on corporations and public institutions to “invest in shared open infrastructure for AI defense — datasets, evaluation frameworks, attack simulators and red-teaming tools — much as past generations invested in open source software,” per Nvidia's statement.

Frequently Asked Questions

What are open-weight AI models?

Open-weight AI models are models whose internal parameter weights are accessible, allowing organizations to deploy and execute them on their own private servers or cloud hardware instead of relying exclusively on closed proprietary API services.

Why are companies switching from proprietary models to open-weight AI?

Businesses are seeking to curb rapidly rising technology expenses. High-volume usage of proprietary models from vendors like OpenAI and Anthropic can lead to sharp budget increases, whereas open-weight models allow companies to run workloads on their own hardware without recurring per-query vendor fees.

How fast is enterprise interest in open-weight AI growing?

According to AlphaSense data reported by the Financial Times, executive mentions of “open weight” and “open source” models during corporate earnings calls and investor presentations increased sixfold in August and September compared to the same two-month period in 2025.

What challenges come with deploying open-weight AI?

While open-weight models offer cost predictability, operational flexibility, and data control, organizations must assume the responsibility and expense of managing the computing infrastructure, hardware, and operational environments required to run them.

Sources & Further Reading

• PYMNTS