AI Security Alliance Formation and Infrastructure Expansion Race Intensifies

인프라/플랫폼 | Tue Jul 28 2026 00:00:00 GMT+0000 (Coordinated Universal Time) | 7 sources

Nvidia and Microsoft formed an open source AI security alliance while SSI secured a compute partnership, as data center power issues emerged.

Analysis

[Nvidia] launched the Open Secure AI Alliance ^[1]

Co-founded with Microsoft
SpaceX
IBM
and others
Building and sharing open source AI security tools
OpenAI
Google
and Anthropic did not participate
Palantir
Linux Foundation
Cloudflare
Dell
and Cisco joined as founding members

[Microsoft] unveiled the MAI-Cyber-1-Flash model and Project Perception platform ^[2]^[4]

Scored 96% on the CyberGYM benchmark
12-point lead over Anthropic Mythos
Automates red
blue
and green team agents
Preview available on November 3

[Safe Superintelligence] signed a long-term compute partnership with Nvidia ^[6]

Secured access to the Vera Rubin GPU platform
10x expansion of compute resources
Approximately $5 billion investment
Raised $7 billion total at a $32 billion valuation

[Satya Nadella] warned that companies relying on a single AI cannot survive ^[3]

Emphasized the need to adopt AI gateway infrastructure
Recommended separating harness from model
Advised against built-in tools like Claude Code and ChatGPT Codex
Urged retaining metadata for training proprietary open models

[PJM Interconnection] experienced grid instability due to simultaneous data center drop-off ^[7]

3GW load lost within 30 seconds due to transmission line fault
Voltage spike from Northern Virginia to Chicago
Took 11 minutes to stabilize
Approximately 3% of PJM's total demand

[TechCrunch Disrupt 2026] will discuss AI power infrastructure on the Smart Systems Stage ^[5]

Commonwealth Fusion Systems
Helion
and others participating
Discussing commercial fusion grid integration approaches
Highlighting aging grid modernization and investment flows
Addressing data center power procurement bottlenecks

Sources