Meta says its Llama models were downloaded ~350M times, are used by AT&T and other companies, and usage via cloud providers more than doubled from May to July
Meta's (META.O) Llama artificial intelligence models are being used by companies including Goldman Sachs (GS.N) and AT&T …
Context & Ripple Effects
This adoption update follows Meta’s release of Llama 3.1 models including a 405B version, giving companies a newer model family to evaluate through cloud channels rather than only as self-managed downloads.
The reported enterprise users and cloud-usage growth make distribution, not just model release, the key signal. Later coverage of Llama reaching 1B downloads suggests this was an early marker of a widening installed base.
First-order effects
- Meta gains evidence that Llama is reaching both enterprise users, including AT&T and Goldman Sachs, and customers accessing models through cloud providers.
- Cloud providers serving Llama see higher model usage, while Meta can point to roughly 350M downloads as a measure of developer and enterprise reach.
Second-order effects
- Enterprise buyers gain more leverage to compare Llama deployment options across cloud providers and against proprietary-model offerings, increasing pressure on vendors to compete on access, integration and operating cost.
- The cloud channel becomes more consequential to Llama’s adoption: growth in usage there can matter more commercially than downloads alone because it reflects ongoing workloads.
Third-order effects
- If enterprise use continues to expand, model competition will increasingly be decided by distribution through clouds and enterprise integrations, alongside benchmark performance.
- A larger deployed base can make Llama a more durable platform for specialized implementations; the later extension to US national-security applications shows how access can broaden into additional institutional markets, subject to customer and policy constraints.
The trend: AI model providers are turning broad model availability into a distribution race across cloud platforms, enterprises and institutional users.