When Sber's research team quietly dropped two open-weight AI models on Hugging Face — including a 702-billion-parameter MoE system that reportedly outperforms DeepSeek and Qwen on math benchmarks — Startups and Ventures had the breakdown ready within hours. That kind of speed and technical specificity is what separates this channel from generic tech aggregators, and it signals exactly what kind of operation this is.
Running under the handle @technology on Telegram, the channel has accumulated just over 3 million subscribers, making it one of the larger English-language tech feeds on the platform. The content rhythm runs at roughly 2-3 posts per day, mixing hard AI news, space coverage, developer tooling drama, and the occasional hardware curiosity — like Amida's 100-piece NASA tribute watch, which felt slightly out of place but at least came with real specs and context.
The editorial focus is clearly weighted toward AI infrastructure and developer ecosystems. Recent posts covered OpenAI's policy paper on restructuring labor taxation and public wealth funds in response to automation, Anthropic disabling third-party Claude subscription workarounds, a leaked Claude Code roadmap, Netflix open-sourcing its VOID video inpainting model, and Apple quietly banning vibe-coding apps — only to watch developers migrate the functionality into iMessage. These aren't surface-level headlines. Each post includes enough technical detail to be genuinely useful: model parameters, benchmark comparisons, licensing terms, direct links to Hugging Face repos and official papers.
The writing style is tight and functional. No fluff, no hot takes, no opinion columns. Posts read more like well-edited briefings than commentary. That's a deliberate choice, and it works for the audience this channel is clearly serving — developers, AI researchers, startup operators, and investors who want signal without noise. The channel also covers space and hardware news, though less consistently. The Artemis II launch got solid coverage, as did the Baidu robotaxi mass-stall incident, which was framed usefully alongside the earlier Waymo outage for context.
Where the channel falls short is depth. A post noting that Claude Code now refuses to analyze its own source code is interesting — but a single line with no follow-up leaves readers wanting more. The same goes for the Dubai firefighting tech post, which was literally just a video with no text at all. For a channel this size, occasional lazy posts stand out more than they would elsewhere.
There's also a companion community at @startupdis for those who want to discuss the content, and advertising is sold through @strategy — worth noting for anyone evaluating the channel's commercial positioning. The audience size means sponsored posts appear regularly, though they haven't overwhelmed the editorial feed in recent weeks.
For founders tracking AI tooling, developers monitoring open-source releases, or investors watching how major labs are positioning their products, this channel delivers consistent, technically credible coverage. It won't give you analysis or opinion, but as a daily briefing layer for what's actually shipping in the AI and startup world, it earns its place in the feed.
S
