Key Highlights
- Alexandr Wang, Meta’s superintelligence chief, informed staff that the company’s forthcoming Watermelon AI model has achieved performance parity with OpenAI’s GPT-5.5.
- The Watermelon model utilizes approximately ten times more computational resources than Avocado, Meta’s internal designation for the Muse Spark model launched recently.
- META shares declined 4.90% in trading following the internal disclosure, contradicting the optimistic AI development news.
- The social media giant has revised its 2026 AI infrastructure spending forecast to $125–$145 billion, increased from the previous $115–$135 billion projection.
- An upcoming Muse Spark refresh will deliver substantial improvements in coding performance and agentic functionality, according to Wang.
Shares of Meta (META) experienced a 4.90% decline on Friday, despite emerging reports of significant progress in the company’s artificial intelligence initiatives. During an internal company meeting, Alexandr Wang, who heads Meta’s superintelligence division, informed staff members that the tech giant’s forthcoming AI system, internally referred to as Watermelon, has achieved performance levels comparable to OpenAI’s leading GPT-5.5 model — a development that would represent a significant breakthrough in Meta’s competitive AI strategy.
Wang referenced various AI performance benchmarks to support his assertion, although the specific evaluation metrics used remain undisclosed.
The Watermelon system represents Meta’s successive generation following Avocado — the company’s internal code name for Muse Spark, which launched in April. While Muse Spark demonstrated competitive performance on industry benchmarks upon release, it remained behind the leading models from OpenAI and Anthropic.
According to Wang, Watermelon requires computational resources roughly ten times greater than those used by Avocado. Increased computational power generally correlates with enhanced model capabilities, though it simultaneously drives up operational expenses.
Enhanced Coding and Autonomous Capabilities Coming
Wang also shared some developments publicly. Through a post on X, he announced that an enhanced version of the existing Muse Spark model is approaching release, featuring substantial upgrades to coding proficiency and agentic functionalities.
When questioned by a user about when Meta would deliver a coding-focused model competitive with Anthropic’s Claude Opus, Wang indicated it would arrive “pretty soon” and hinted that users would appreciate what the company is “cooking.”
Meta has invested considerable resources over recent years attempting to narrow the competitive distance with OpenAI, Google, and Anthropic. Substantial expenditures on processing chips, computing infrastructure, and premier AI researchers have not yet resulted in market dominance — at least not in publicly available products.
Zuckerberg brought Wang into the organization last year to spearhead what is now called Meta Superintelligence Labs. Wang manages an elite research division known internally as TBD, in addition to overseeing the company’s comprehensive AI and hardware initiatives.
Capital Expenditures Continue Rising
Meta has reportedly extended compensation packages worth hundreds of millions of dollars to attract leading AI researchers. The financial commitments extend far beyond talent acquisition.
The corporation recently informed shareholders that it anticipates capital expenditures between $125 billion and $145 billion during 2026 for processors, computing facilities, and supporting infrastructure. This represents an increase from the previously announced range of $115–$135 billion, attributed to escalating component prices and expanded data center construction plans.
For perspective, OpenAI launched GPT-5.5 in April. The company subsequently unveiled GPT-5.6 late last month — characterized as its most capable system to date — though widespread access has been limited, reportedly due to requests from US government officials.
Wang’s performance claims have not received independent confirmation, and Meta representatives declined to provide comment. OpenAI similarly did not respond to inquiries.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
