DAILY NEWS

Stay Ahead, Stay Informed – Every Day

Advertisement
WORLD
The Strait of Hormuz is unblocked and Japanese ships stranded in the Persian Gulf begin to pass through | International | Central News Agency CNA
jackminion 0



Please agree to our privacy policy to enable news listening. (Central News Agency, Tokyo, 19th Comprehensive Foreign Report) According to the ship tracking platform MarineTraffic, as the three-and-a-half-month blockade of the Strait of Hormuz has been lifted, Japanese-related ships have begun to leave the Persian Gulf and sail through the strait. Japan’s “Yomiuri Shimbun” reported that after the United States and Iran signed a 14-point memorandum on the 17th, it is expected that a large number of ships originally stranded in the Persian Gulf will begin to pass through the Strait of Hormuz. The report cited MarineTraffic data and pointed out that a Saudi oil tanker destined for the Kii area of ​​Kagoshima City, Japan, restarted and sent positioning information in the Gulf of Oman on the east side of the Strait of Hormuz on the 18th. The ship was shut down in the Persian Gulf in mid-April and was presumably reopened after passing through the strait and sailing out of the Gulf. In addition, the oil tanker TENZAN, which was originally anchored off the coast of the United Arab Emirates and operated by a Japanese company, has also begun to move towards the Strait of Hormuz and entered the strait on the evening of the 18th. The report quoted relevant sources from the Japanese government as saying that there were Japanese crew members on board. Japan’s Ministry of Land, Infrastructure, Transport and Tourism and other units stated that as of the 18th, there were 38 Japanese-related ships stranded in the Persian Gulf, with a total of about 900 people trapped. Chief Cabinet Secretary Minoru Kihara said at a press conference this morning: “In order to allow all ships to pass through the Strait of Hormuz as soon as possible, we will continue to devote all our diplomatic efforts.” (Compiled by: Li Jing / Verified by: Shi Shi) 1150619 Support the Central News Agency’s Choice and Facts Station Together, every donation you make is the power to protect press freedom. For small donations, download the Central News Agency’s “First-hand News” APP to get the latest news in real time. The text, pictures, and audio and video on this website may not be reproduced, publicly broadcast, or publicly transmitted and used without authorization.



Source link

WORLD
Uber takes first place in Japan’s ride-hailing app market and invests 63.2 billion in the next five years | International | Central News Agency CNA
jackminion 0



2026/6/14 16:52 (updated at 6/14 17:32) Please agree to our privacy policy to enable the news listening function. Schematic diagram. (Picture taken from Unsplash Gallery) (Central News Agency, New York, 14th, comprehensive foreign news reports) Andrew Macdonald, President and Chief Operating Officer of Uber, an American transportation network company, revealed that in Japan’s ride-hailing service market, Uber surpassed competitors such as GO for the first time in April, taking the top spot in market share. Japan’s Kyodo News reported that McDonald said this was the result of counting credit card payment data. At the same time, he also announced that he will invest more than US$2 billion (approximately NT$63.2 billion) in the next five years to further promote business expansion. In an exclusive interview with Kyodo News in New York, McDonald pointed out that Uber’s market share is continuing to expand by attracting tourists to Japan and strengthening cooperation with local taxi companies. He emphasized that Japan is an “extremely important market” and pointed out that “Uber’s goal is to become indispensable to Japanese consumers, whether in ride-hailing or delivery services.” He also added that he hopes to help Japan cope with social issues such as aging by improving the convenience of the App, thereby contributing to the Japanese economy. The US$2 billion investment will be used to increase the number of drivers, advertising and promotional activities. In view of the inconvenient “traffic blank areas” in the suburbs, Uber will assist local governments in introducing “self-used minibuses to participate in paid passenger transportation services.” According to Japan’s Ministry of Land, Infrastructure, Transport and Tourism’s “Paid Passenger Transport Service Manual for Private Passenger Vehicles”, “traffic blank areas” refer to areas that lack public transportation or are extremely difficult to use existing transportation, and are facing development difficulties in regional transportation. “Paid passenger transportation service for private use passenger vehicles” refers to a transportation mode implemented by municipal governments or non-profit organizations as the main body after joint agreement between local relevant parties and obtaining registration permission in accordance with the “Road Transportation Act” when existing bus and taxi operators are unable to provide transportation services. In addition, Uber also plans to cooperate with Nissan and other companies in the second half of this year to trial operate self-driving taxis in Tokyo. (Compiled by: Li Jing / Verified by: Chen Zhengjian) 1150614 Support the Central News Agency’s choice to stand with the facts. Every donation you make is a force to protect press freedom. Small-amount sponsorship downloads the Central News Agency’s “First-hand News” APP to get the latest news in real time. The text, pictures and audio and video of this website may not be reproduced, publicly broadcast or publicly transmitted and used without authorization.



Source link

TECH & AI
How I built a 6-node 12-GPU on-prem AI cluster running 1000+ agents
jackminion 0


TL;DR — 6 machines, 12 GPUs, 1,000+ concurrent agents, P95 18 ms, voice

Why I built this

I’m Franck. Toulouse, France. Over 3 years I paid roughly €280,000 to Azure + OpenAI before doing the math properly:

Latency: 1.2s voice round-trip — incompatible with the voice-first UX I wanted.

Compliance: customer data on US servers. Not GDPR-native, just GDPR-compliant-on-paper.

Quotas: random throttling at the worst times.

Lock-in: Azure outage = my product offline.

I decided to rebuild everything on-prem. This is the result.

The cluster

6 machines, 3 tiers, 12 GPUs total,

Tier 1 — GPU compute (heavy inference)

M1 “La Créatrice” — Ryzen 5700X3D, 6× RTX 3080+, 46 GB RAM. Primary LLM node, runs qwen3.5-9b, qwen3.5-35b-a3b, deepseek-r1, the Claude 4.5/4.6 distillations, and the Whisper CUDA pipeline.

M2 “Le Forge” — multi-GPU NVIDIA, secondary inference, failover from M1 in 1.3s.

Tier 2 — CPU/RAM (orchestration, memory)

M3 “Le Cerveau” — high-RAM CPU node. PostgreSQL + Redis + Pinecone. Runs the orchestrator, the 3-quorum consensus engine (M1+M2+M3), and the analytics/monitoring agents.

Tier 3 — production / work

M4 “Bridge Windows” — Windows 11, 2 GPUs, trading bot live.

M5 “Interface Relay” — Linux i5-6500, 15 GB RAM. Dev interface, 15+ MCP servers, Claude Code.

M6 “Mobile Ops” — laptop. SSH + VPN. Client demos and on-site ops.

The 9 layers I added on top of Ubuntu

L9 — Vocal / conversational (Whisper CUDA STT, Piper TTS, wake word, 50+ languages)
L8 — Multi-agent orchestration (MCP-native, consensus engine)
L7 — Trading consensus engine (multi-model voting GPT/Gemini/Claude)
L6 — Browser + web automation (Chrome DevTools Protocol)
L5 — MCP tool registry (88+ handlers)
L4 — GPU cluster management (Docker Swarm, failover
L3 — Domino pipeline engine (835 chains)
L2 — systemd service layer (98 units)
L1 — Linux boot integration (GRUB hooks, ZRAM, kernel params)

Real numbers

Metric
Value

Concurrent agents
1,000+

P95 latency (cluster internal)
18 ms

Voice pipeline end-to-end

Aggregate throughput
67 tok/s

Python lines
280,741

Public repos
44 (all MIT)

Cost comparison (1M tokens/day, team of 10)

Provider
€/month
P95
Concurrent agents
Data residency

Azure OpenAI
1,500
800ms-3s
~20
US

AWS Bedrock
1,800
700ms-2.5s
~15
US

Mistral Cloud
800
400-800ms
~30
EU

JARVIS OS
0
18 ms
1,000+
Air-gapped

For a 50K€ turn-key deployment, break-even vs Azure is 7 months, and the marginal cost is zero after that.

What I sell now

JARVIS OS turn-key — 20K€ to 250K€ depending on scope.

62 PDF trainings — from €39, 293h of content based on production code (+48 private).

IA infra audit — €1,500, report in 48h.

1-to-1 mentorship — €250/h.

Fractional CTO — TJM €1,000-1,150 / CDI €85-95K. Toulouse / remote.

Honest weaknesses

Consensus voting is empirical. No formal verification of the agreement function.

Tier-2 failure (M3 down) is the weakest scenario — orchestrator dies, cluster keeps inferring but loses persistent memory.

MCP protocol bet — if Anthropic deprecates parts of MCP, I have 88 handlers to refactor.

kWh-per-token efficiency — cloud probably wins on aggregate watts/token, on-prem wins on marginal cost.

Links



Source link