Internet security company Palo Alto Networks’ CEO Nikesh Arora has emphasized the necessity for a 10-fold increase in the computational efficiency of artificial intelligence (AI) models and a significant reduction in the cost of using AI for the widespread adoption of AI in the enterprise sector.
In an interview with CNBC on June 9, Arora stated that the global demand for AI computation is nearly limitless, but the existing computing power is unable to simultaneously meet the needs of both consumer and enterprise markets. He pointed out that in the next 6 to 12 months, the focus of the market discussions will shift from how intelligent the models are to the speed of deploying these models.
Arora highlighted that the current cost of using AI in enterprises is still relatively high, and the efficiency of “tokens” (the basic unit for AI model processing and billing related to text) must increase by 5 to 8 times in the next 12 months and 10 times over the following year.
“We need to see a decrease in AI prices,” he said. “At the current prices, it is challenging for enterprises to make a firm decision to fully embrace AI. We do not need AI model companies to add more barriers in terms of pricing and deployment.”
With the rapid proliferation of AI, Palo Alto Networks has emerged as one of the biggest beneficiaries. Arora mentioned that the company’s stock price has surged by nearly 77% so far this year.
Palo Alto Networks announced its latest financial results on June 2, revealing a 31% year-over-year revenue increase, reaching approximately $3 billion, including a contribution of $388 million in revenue from the acquisitions of CyberArk and Chronosphere last year.
In a press release accompanying the financial report, Arora stated that the latest advancements in AI technology have led to increased focus on network security within enterprises and have redefined the industry development direction for the coming years.
Other major cybersecurity companies have also reaped benefits from the AI boom, with CrowdStrike seeing a nearly 60% increase in its stock price this year.
On July 8, Elon Musk’s xAI (now known as SpaceXAI) unveiled the latest chatbot model, Grok 4.5, with pricing set at $2 for every 1 million input tokens and $6 for every 1 million output tokens. The previous generation, Grok 4.3, had pricing of $1.25 and $2.5 respectively.
OpenAI’s latest GPT-5.6 model charges between $1 to $10 for every 1 million input tokens and $6 to $45 for every 1 million output tokens, depending on the version.
The pricing for Anthropic’s newest model, Claude Fable 5, is $10 per 1 million input tokens and $50 per 1 million output tokens, whereas some of the company’s older models are priced at half of those costs.