Meta Shifts to Open-Source AI Strategy as CoreWeave Targets $10B Revenue in 2026
- Sources
- 2
- Claims verified
- 2 of 3
What happened
As of April 2026, Meta Platforms and CoreWeave have emerged as pivotal players in the artificial intelligence sector. Meta is shifting its strategy by planning to release versions of its next-generation AI models as open source, aiming to foster a broader developer ecosystem while competing with proprietary systems from OpenAI and Anthropic. These models, developed under the leadership of Alexandr Wang following a $15 billion deal with Scale AI, will keep safety-critical components and the largest models proprietary. Simultaneously, CoreWeave is positioned as a near pure-play AI infrastructure provider, with revenue projected to exceed $10 billion in 2026. While CoreWeave demonstrates explosive growth, it faces risks including heavy capital expenditures and a high revenue concentration from Microsoft. Meta continues to integrate AI across its social media suite, reaching over 1 billion monthly active users, while utilizing its Llama models to automate advertising processes and enhance user engagement.
Key points
- CoreWeave is projected to generate over $10 billion in revenue in 2026, up from minimal sales in 2022, though it remains unprofitable with heavy debt.1
- Meta Platforms plans to release parts of its next-generation AI models as open source to compete with proprietary systems from OpenAI and Anthropic.2
- Microsoft accounted for 67% of CoreWeave's total revenue in 2025, representing a significant concentration risk for the infrastructure provider.1
- Meta's new AI models are the first developed under Alexandr Wang, following a $15 billion deal with Scale AI in 2025.2
- Meta has integrated AI across its apps, reaching 1 billion monthly active users and using Llama models to automate advertising processes.1
Why it matters
CoreWeave's 'pure-play' status is a double-edged sword; while it captures AI infrastructure growth, its 67% revenue reliance on Microsoft makes it highly vulnerable to changes in a single client's capex strategy.
Meta's open-source pivot is a strategic attempt to commoditize AI models, potentially eroding the pricing power of proprietary model developers like OpenAI while strengthening Meta's own ad-targeting ecosystem.
The $15 billion Scale AI deal and the appointment of Alexandr Wang signal that Meta is prioritizing high-quality data and model training efficiency over just raw compute power.
In context
| Round | Date | Post-money valuation | Raised | Led by |
|---|---|---|---|---|
| Series B extension | May 31, 2023 | $2.0B | $200M | Magnetar Capital |
| Series C | May 1, 2024 | $19B | $1.1B | Coatue |
Updated
Fact-check
- Meta Plans to Release Open-Source Versions of Upcoming Advanced AI ModelsMultiple outlets (Axios, The Decoder, Economic Times) confirm Meta is preparing new models under Alexandr Wang with a hybrid open-source strategy to manage safety risks.Verified
- CoreWeave and Meta remain top picks / See Significant Investor Interest in 2026Market commentary from The Motley Fool and Yahoo Finance confirms these companies are top subjects for AI investors, citing specific revenue growth ($5.1B in 2025) and stock performance (~62% annual surge for CoreWeave).Partly verified
- Analysts are tracking these stocks closely as they navigate the evolving IT/AI sectorActive coverage from Seeking Alpha, Motley Fool, and Yahoo Finance demonstrates that analysts are indeed tracking these stocks closely as the sector evolves.Verified
Sources
- [1]The Decoder (opens in a new tab)Meta Plans to Open-Source Parts of Its New AI ModelsAccording to this source, Meta is shifting its AI strategy to release open-source versions of its next-generation models to compete with proprietary enterprise systems.High factualityBias 0 from −6 (left) to +6 (right)
- [2]The Motley Fool (opens in a new tab)AI Stocks to WatchReported that CoreWeave is projected to reach $10 billion in revenue by 2026 despite significant risks from high customer concentration and lack of profitability.High factualityBias 0 from −6 (left) to +6 (right)
L–R: source bias
