Key Highlights
- Meta Platforms (META) shares declined 5% following announcement of expanded AI infrastructure partnership.
- Firmus Technologies, based in Australia, will provide GPU computing power across Southeast Asia for Meta.
- Partnership extends Meta’s current Nvidia GB300 NVL72 deployment at Firmus’s Melbourne facility.
- Meta’s Q2 capital expenditures reached $31.1 billion with free cash flow plummeting 91%.
- Analysts maintain a Strong Buy consensus on META stock with price targets suggesting 12% upward potential.
Shares of Meta Platforms (META) experienced a 5% decline after the social media giant announced an expanded AI computing collaboration with Firmus Technologies, an Australian infrastructure provider.
The partnership establishes Firmus as a key supplier of graphics processing unit computing resources for Meta. These resources will originate from a network of AI facilities currently under construction throughout Southeast Asia.
Meta intends to leverage this expanded capacity for artificial intelligence research initiatives, large-language model development, and machine learning training operations. Neither company revealed the financial parameters of the arrangement.
Firmus specializes in constructing and managing AI-optimized data centers designed for intensive computational tasks. The company provides high-performance computing infrastructure, cloud services, and AI-as-a-service solutions to enterprise and government clients.
Extending Current Collaboration
The newly announced agreement strengthens the existing relationship between Firmus and Meta that’s already operational in Australia. Meta currently utilizes Nvidia GB300 NVL72 infrastructure housed within Firmus’s Melbourne data facility.
According to Firmus, this Melbourne installation represents the most extensive Nvidia Blackwell Ultra system deployment south of the equator. The upcoming Southeast Asian facilities will operate using Nvidia’s DSX AI Factory architecture.
This architecture will integrate with Firmus’s proprietary HyperCube thermal management system. Effective cooling solutions are critical for AI chip operations, as these processors produce substantial heat during intensive computational workloads.
Firmus’s liquid cooling approach is engineered to reduce power consumption and minimize operational expenses. This infrastructure could enable Meta to efficiently scale its artificial intelligence training requirements.
Meta’s Aggressive AI Investment Strategy
This Firmus partnership arrives amid Meta’s continued escalation of capital investments in AI-related infrastructure. During the second quarter, Meta’s capital expenditures totaled $31.1 billion.
Free cash flow collapsed 91% in the same period, dropping to just $784 million. Meta has indicated potential capital spending of up to $145 billion throughout the current fiscal year.
For Firmus, this agreement represents a significant milestone in its Asia-Pacific growth strategy. Securing Meta as a major customer strengthens credibility for its infrastructure offerings.
The partnership also provides Firmus with more stable, recurring revenue streams from GPU capacity commitments. This financial stability could prove beneficial as the company advances toward its anticipated $5 billion initial public offering on the Australian Securities Exchange.
Should this IPO proceed successfully, it would represent Australia’s second-largest market debut on record. Firmus develops and operates specialized AI infrastructure facilities that deliver large-scale computing capabilities to external customers.
Wall Street analysts continue to maintain a Strong Buy consensus recommendation on META stock. This collective rating reflects 39 Buy recommendations and six Hold recommendations issued within the last three months.
The consensus price target among Wall Street analysts stands at $798.74 per share. This target represents potential upside of approximately 12% from META’s current trading levels.
Shares of Nvidia (NVDA) gained 2% following the announcement, as Nvidia’s semiconductor technology remains fundamental to the AI infrastructure strategies of both Meta and Firmus.





