
Hacker Newsroom for 15 November: Nvidia GPU Financing, Lovable Data Exposure, ggml 1.0.0
15 Nov 2026 · 3:27
Hacker Newsroom for 15 November recaps major Hacker News stories, moving through nvidia gpu financing, lovable data exposure, ggml 1.0.0.
Chapters
- 00:00:00 — Intro
- 00:00:20 — Nvidia GPU Financing
- 00:01:19 — Lovable Data Exposure
- 00:02:19 — ggml 1.0.0
- 00:03:20 — Closing
1. Nvidia GPU Financing
Nvidia is facing allegations that it financed AI startups and smaller cloud providers that then bought its GPUs, according to a TechPowerUp summary of Wall Street Journal reporting; the deals could help sustain demand and absorb large TSMC orders even when some buyers could not afford the hardware. The report says CFO Colette Kress pushed to scale back these arrangements, but it presents allegations rather than proof that Nvidia engineered a GPU shortage.
2. Lovable Data Exposure
A security investigation into Lovable apps found that thousands of sites may have exposed data through misconfigured Supabase databases. In a sample of 159,041 projects, 24,664 returned database records without login, and further checks found email-containing records across 7,403 sites; the researchers caution that this sample does not establish a platform-wide leak rate.
3. ggml 1.0.0
Ggml 1.0.0 updates the open-source tensor library used by projects like llama.cpp, with a cleaner public API, RPC version 9, configurable YaRN truncation, and a TRI backward pass. The release also brings performance and backend work across a wide range of hardware, including faster quantized matrix operations, mixture-of-experts, attention, and scheduling.
That's the digest.