Google has reportedly told employees that the company must double its capacity every six months to keep up with soaring demand for artificial intelligence tools. The guidance was shared by Amin Vahdat, Google’s Vice President of AI and Infrastructure, during a recent all-hands meeting, according to a CNBC report.
Vahdat outlined the scale of what Google is facing, telling staff that the company needs to achieve “the next 1000x in 4-5 years,” a target that reflects how quickly AI workloads are growing across the industry. His message highlighted a future defined by enormous infrastructure expansion alongside strict demands for efficiency, cost control, and energy use.
Scaling capacity to meet rising AI demand
During the meeting, Vahdat explained that achieving this level of growth by the end of the decade would require Google to accelerate its progress significantly. He stressed that the company must deliver these increases at “essentially the same cost and increasingly, the same power, the same energy level.” This means Google will need to find ways to support vastly larger AI models and workloads without multiplying its operating expenses or energy consumption.
The pressure to expand capacity reflects the global surge in AI development. As more businesses adopt generative AI systems, cloud providers like Google are being forced to build out infrastructure while keeping services reliable and affordable rapidly.
Building hardware and infrastructure to support growth
Google’s roadmap to meet these needs involves both expanding its data centre footprint and increasing the use of its in-house hardware. The company has been developing custom chips, including its Tensor Processing Units, to reduce dependence on external suppliers.
The latest version, Google’s seventh-generation TPU, known as Ironwood, is designed to boost efficiency significantly. The company says Ironwood delivers a 30-fold improvement in power efficiency compared with models released in 2018, highlighting the level of innovation required to meet Vahdat’s targets.
At the same time, shortages of third-party chips continue to affect the industry. Many of Nvidia’s AI processors are currently marked as sold out, as reported by The Verge, limiting availability for major tech firms and slowing deployment of new AI features. Google, like many others, has been impacted by these supply constraints as demand continues to spike.
Leadership warns of tougher years ahead
Alongside Vahdat’s comments, Google CEO Sundar Pichai has also cautioned employees that the coming years will be demanding. He recently described 2026 as likely to be “intense” due to heavy AI competition and the growing need for compute resources.
Although the company has publicly acknowledged concerns about an AI bubble, Pichai believes that failing to invest sufficiently in AI would pose an even greater risk to Google’s long-term position. His statements reinforce the message that the company must continue scaling aggressively despite economic pressure and power limitations.
As the AI race accelerates, Google’s leadership appears to be preparing staff for a decade marked by rapid expansion, technological milestones, and ongoing challenges in balancing efficiency with unprecedented computational demands.
