Enterprise Software Stocks Positioned to Lower AI Token Costs

Investors spent the last two years flocking to semiconductor firms while ignoring enterprise software companies. Many feared that large language models would eventually make traditional software tools obsolete. Recent analysis suggests that certain software providers are actually essential for making artificial intelligence affordable and efficient for major corporations.

Software Solutions for High AI Operational Expenses

The cost of deploying artificial intelligence agents at a massive scale is exceeding original corporate budgets. Piper Sandler analysts recently informed clients that five specific infrastructure software companies are uniquely positioned to address this financial challenge. The list includes MongoDB, Snowflake, Elastic, GitLab, and Atlassian. These firms manage the foundational data that modern AI systems require to function accurately.

A token represents a fragment of text used by AI models to process information. Because AI providers charge customers based on the number of tokens processed, costs escalate quickly as usage grows. The argument from analysts is that the proprietary data stored within these software platforms acts as a context layer. This layer makes AI models more precise while significantly lowering the number of tokens required for each query.

Early data indicates that providing clean organizational context directly to an AI agent can reduce token usage by 50% to 75%. This efficiency stems from a simple mechanical reality of the technology. When an AI receives organized and relevant data, it reaches a conclusion faster and uses fewer computational resources. This allows businesses to expand their use of automation without a linear increase in their monthly technology bills.

The Shift in Corporate AI Spending Patterns

The financial landscape for AI changed significantly throughout 2026 as businesses adjusted their strategies. Even though the price per individual token decreased for newer models, total bills continued to rise for most enterprises. This occurred because newer, more advanced reasoning models consume far more tokens to think through complex problems. A single request that once used hundreds of tokens now requires tens of thousands to complete.

Corporate behavior is shifting away from a strategy of maximum token consumption. Instead of using the most powerful model for every task, companies now use model routing. This process directs simple tasks to inexpensive, small models while saving high-cost models for complex reasoning. Software providers like Snowflake have even updated their billing systems to help customers track these specific costs separately from standard data processing.

Leveraging Context as a Competitive Edge

The strategic value of these software firms lies in their role as the context layer for the enterprise. Model providers like OpenAI or Anthropic offer the raw intelligence, but they lack the specific internal data of a corporation. Companies that own the data repositories become the gatekeepers for efficient AI. This relationship creates a defensive moat that is difficult for pure AI startups to penetrate without access to that same historical data.

Transitioning to Consumption Based Revenue Models

One of the most significant changes in the software industry is the move away from per-seat licensing. In the past, software companies made money based on the number of employees using the tool. There is a growing fear that AI will eventually reduce human headcount, which would hurt revenue for companies using the old model. To counter this, many firms are adopting consumption based pricing.

Under a consumption model, a vendor earns more revenue as their customers use more AI services. If a company deploys more automated agents that perform more queries, the software provider receives a larger payment. This happens even if the total number of human employees stays the same or decreases. This shift provides a new growth path that is not tied to hiring trends or corporate headcount.

Three Requirements for Growth

For this new revenue thesis to succeed, three specific conditions must remain true. First, large enterprises must continue to deploy AI agents at a steady pace. If companies pause their AI rollouts due to security or privacy concerns, the expected revenue growth will stall. The momentum of adoption is the primary driver of the consumption based financial model.

Second, the context layer must remain difficult for AI model creators to replicate. If a company like Google or Microsoft can easily build its own retrieval tools that bypass third party software, the value of firms like MongoDB could diminish. Finally, the revenue gained from AI consumption must be large enough to offset any losses from traditional software licenses. This balance is the key metric that investors are watching during current earnings cycles.

Confirming the Trend Through Market Feedback

Analysts have confirmed through discussions with management teams and partners that organizations are actively seeking software to improve AI efficiency. The demand for cost-saving measures is high because current AI spending is viewed as unsustainable for many departments. By making AI cheaper to run, these software companies are essentially securing their own place in the future corporate tech stack.

Financial Performance and Market Realities

The five companies identified by analysts have seen very different results in the stock market. MongoDB has performed well, seeing its market capitalization reach nearly $28 billion recently. This suggests that some investors have already begun to price in the potential for AI driven growth. On the other hand, Elastic has seen its share price struggle, leading some firms to lower their price targets despite maintaining positive ratings.

GitLab has emerged as one of the weaker performers among the group. Many analysts currently maintain a hold rating on the stock, as it remains heavily dependent on developer seat counts. Atlassian also trades well below its historical targets, though some investment banks maintain that the stock is currently undervalued by as much as 33%. These discrepancies show that the market is still deciding which companies will truly benefit from the AI shift.

The Broader Software Industry Slump

The focus on these five stocks comes at a time when the broader software sector is underperforming. The S&P 500 software industry index is down more than 25% from its peak in late 2023. While the general S&P 500 index has seen gains of around 10% this year, many software funds have dropped by double digits. This creates a challenging environment for any company trying to prove its long term value.

Investment firms like Morgan Stanley suggest that the negative sentiment surrounding software has gone too far. They argue that once AI model providers stop subsidizing their services and raise prices, the need for efficient software will become even more obvious. The current market dip may represent a period of transition where the old ways of valuing software no longer apply, but the new ways are not yet fully understood.

Evaluating Risks and Future Outlook

There are notable risks to the idea that software stocks will be saved by AI efficiency. The primary threat is that AI labs will build their own memory and data retrieval features directly into their platforms. If a model can remember everything a company needs without a third party database, the need for a context layer vanishes. Some major AI labs have already introduced features that compete directly with these software firms.

Timing also remains a significant concern for investors. Analysts are describing a window of opportunity that has not yet appeared in official financial reports. None of these companies currently report their AI context revenue as a separate item. This means that for now, the entire investment thesis is based on estimates and early pilot programs rather than hard financial data.

The upcoming earnings season will be a critical turning point for these stocks. Investors will be looking for specific mentions of net revenue retention and AI driven consumption. If existing customers are spending more money as they deploy more AI agents, it will prove that the software layer is a vital part of the ecosystem. If those numbers remain flat, it may indicate that the software sector still has a long road to recovery.