May 02 AI Daily — Pentagon AI Contracts, Big Tech AI Revenue Surges

1. Pentagon Signs AI Deployment Contracts for Classified Networks with Major AI Firms (Excluding Anthropic)

The US Department of Defense has signed contracts with 7 major AI companies, including SpaceX, OpenAI, Google, NVIDIA, Microsoft, Amazon Web Services, and Oracle, to deploy AI capabilities on classified networks. These agreements aim to transform the US military into an "AI-first" fighting force and enhance decision-making dominance across all warfighting domains. Notably, Anthropic was excluded from the contracts due to disagreements over AI safety guidelines, highlighting the tension between ethical considerations in AI development and the government's demand for control.
Sources: Department of War · The Guardian · 7min.ai  |  🔗 View Original  |  🔗 View Original

---

2. Big Tech AI Revenue Surges Amid Continued Massive Investments

Major Big Tech companies like Meta, Amazon, Microsoft, and Alphabet have dispelled skepticism regarding AI infrastructure investments, proving they are realizing billions in AI revenue. Furthermore, Founders Fund, co-founded by Peter Thiel, raised a new $6 billion fund dedicated to AI startup investments, demonstrating venture capital's strong confidence and aggressive investment stance in the AI industry. This indicates that the AI boom is a solid growth cycle backed by actual customers and revenue.

• Microsoft AI Business: $37 billion annualized revenue (123% year-over-year growth)
• Amazon AWS AI Services: Over $15 billion annualized revenue
• Meta AI-Driven Ad Suite: Over $20 billion annualized revenue
• Alphabet Google Cloud Revenue: Surged 63% driven by generative AI
  Sources: Benzinga · TNW  |  🔗 View Original  |  🔗 View Original  |  🔗 View Original

---

3. Deepening Performance Race in AI Agents and LLMs: Grok 4.3 Launch and GPT-5.5 Cybersecurity Capabilities

xAI launched Grok 4.3, improving cost-efficiency and IQ while offering a 1-million token context window and autonomous execution features. OpenAI's latest model, 'GPT-5.5', was shown in research to perform on par with or slightly exceed Anthropic's 'Claude Mythos Preview' in cybersecurity capabilities across some metrics. Additionally, OpenAI added a 'goal (/goal)' feature to its developer coding tool Codex CLI to autonomously execute long-term objectives, while Google Cloud launched the 'Gemini Enterprise Agent Platform' for developing and deploying AI agents.

• Grok 4.3: Price drastically slashed to $1.25 per input token
• GPT-5.5: Cybersecurity capabilities on par with Claude Mythos Preview
  Sources: xAI · UK AISI · Google Cloud · AI Times · 7min.ai · TLDR AI  |  🔗 View Original  |  🔗 View Original  |  🔗 View Original  |  🔗 View Original  |  🔗 View Original  |  🔗 View Original

---

4. AI Hardware Innovation and Supply Chain Strengthening: Kyocera Develops Advanced Semiconductor Substrates

Kyocera announced the commercialization of a new multilayer ceramic core substrate for advanced semiconductor packaging, such as xPUs and switch ASICs. Built with proprietary fine ceramic materials for high-density wiring and exceptional rigidity, this new product drastically reduces warpage in high-performance semiconductor packages. Meanwhile, Apple discontinued the cheapest Mac Mini model (256GB) and raised the starting price to $799, driven by a surge in demand from engineers looking to run agentic AI tools locally, illustrating how AI infrastructure demand is impacting the consumer desktop market.
Sources: Business Wire · 7min.ai  |  🔗 View Original  |  🔗 View Original

---

5. Strengthening Defenses Against AI Security Threats

Researchers at OX Security discovered a critical command execution vulnerability in the STDIO transport of MCP—the default connection for local tools in AI agents—affecting approximately 200,000 servers. In response to such threats, Anthropic launched the public beta of Claude Security for Enterprise customers, utilizing Claude Opus 4.7 to scan for code vulnerabilities and recommend patches. Additionally, security researchers at Theori demonstrated the potential of LLM-based code auditing tools by finding a critical vulnerability (CVE-2026-31431) in the Linux kernel in just one hour using the AI tool Xint Code AI.
Sources: OX Security · Theori · Anthropic · 7min.ai · TLDR AI  |  🔗 View Original  |  🔗 View Original  |  🔗 View Original  |  🔗 View Original

---

6. AI Cost Efficiency and Shifting Business Models

Uber burned through its entire 2026 AI budget in just four months, primarily spending it on Anthropic's Claude Code. This highlights remarkable productivity gains from AI—with 95% of engineers using AI tools and 70% of committed code being AI-generated—but simultaneously raises questions regarding the cost-efficiency and business models of AI services. Moves such as GitHub Copilot shifting to 'usage-based billing' and Anthropic's Claude Code raising projected costs and signaling a shift toward a pay-as-you-go structure indicate that AI service business models are transitioning from subscription-based to usage-based.
Sources: Briefs.co · 7min.ai  |  🔗 View Original