Piper Sandler称五家软件公司可受益于AI token成本优化趋势

Can Alphabet earnings reignite investor enthusiasm for the Magnificent Seven?
Investing.com -- In a note to clients on Wednesday, Piper Sandler analysts highlighted five infrastructure software companies positioned to help enterprises cut AI token costs by 50% to 75%.
The analysts, led by Rob Owens, named Elastic, GitLab, MongoDB, Snowflake and Atlassian as key beneficiaries of the trend.
Piper Sandler said that as concerns mount around the ROI and pace of AI spending, "a critical window is opening" for its infrastructure coverage.
The firm believes these companies can leverage proprietary enterprise data already residing on their platforms to make AI models "significantly more accurate and efficient while slashing token usage costs," allowing enterprises to scale AI adoption "without a linear increase in costs."
While raw token costs have fallen significantly, with GPT 5.6 output tokens roughly 50% cheaper than GPT 4 output tokens, improved reasoning capabilities in newer models have caused token consumption to "skyrocket," noted Owens.
This is said to have prompted a shift away from "Tokenmaxxing" toward more focused spending strategies, with many enterprises turning to model routing to cut costs.
Piper Sandler said the organizational context and data customers already manage within existing software solutions "can materially improve an agent's accuracy," with early use cases showing 50%-75% reductions in token usage.
Because vendors price these "context layers" based on consumption, the firm sees this as "a compelling incremental growth opportunity that also strengthens longer-term moats."
Piper Sandler said conversations with management teams and channel partners have confirmed that organizations are increasingly looking to software to make AI more efficient.