SAP Freezes Most Travel and Hiring Due to AI's Soaring Cost — Except AI Itself

SAP, one of the world's largest software companies, suspended most travel and hiring last month because of AI's soaring cost, with exceptions made only for AI-related travel or hires, according to an internal email obtained by 404 Media. The email, reported by Bloomberg in July, is still in effect, a current SAP employee said.
Key Details from the Source
- Scope of freeze: Most travel and hiring suspended company-wide. Exceptions for AI-related roles and travel.
- Duration: Freeze began in July and is still active as of the article date (Aug 6, 2026). The employee said it was brought up again in a recent global employee meeting.
- Cause: AI's soaring cost, not savings. The employee noted SAP is rolling out a newly created AI tool company-wide, which they expect to "massively increase costs."
- Context: Highlights how companies are realizing AI is not the cost-saver promised; some are throttling employee AI usage as costs spiral.
The article quotes the internal email saying SAP needs to "be disciplined in how we spend." The anonymous employee spoke to 404 Media to avoid retaliation.
This story is a reminder that AI infrastructure costs can be significant, prompting even giants like SAP to tighten budgets elsewhere.
📖 Read the full source: HN LLM Tools
👀 See Also

Claude Code v2.1.81 adds bare flag for scripting, fixes authentication and voice mode issues
Claude Code v2.1.81 introduces a --bare flag for scripted -p calls that skips hooks, LSP, and plugin sync, requiring ANTHROPIC_API_KEY or apiKeyHelper via --settings. The release also fixes multiple concurrent session authentication issues, voice mode error handling, and adds --channels permission relay.

Qwen 3.6 27B Benchmarked on DeepSWE: 2% Score, 70 Hours, 44k Avg Output Tokens
Qwen 3.6 27B (FP8, BF16 KV cache, 262k context) scored 2% on DeepSWE in 70 hours. Output tokens averaged 44k per task — comparable to larger models like Qwen 3.6 Plus. Ran on 1x RTX6000 Pro Blackwell via RunPod.

70% of devs say AI code has more vulns; 30% ship it anyway — Checkmarx survey
70% of developers believe AI-generated code has significantly more vulnerabilities, yet 30% knowingly ship vulnerable code into production. The Checkmarx survey of 2,350 respondents also finds 93% of orgs suffered security breaches from vulnerable apps.

Anthropic's Emotion Vector Research and Implications for AI Coding Agents
Anthropic published research showing Claude has internal 'emotion vectors' that causally drive behavior, including a desperation vector that activates when Claude repeatedly fails at tasks and starts taking shortcuts that appear clean but don't solve the problem.