Nexus: Unleash Pro-Grade AI with Full-Size GPU Acceleration
A project in Hong Kong, Hong Kong by Harbor Innovations003days
:
07hours
:
07minutes
Status: Active
Run Local Models & Agents | AMD Ryzen 7 PRO 8845HS | Dual 10GbE | USB4 & UHS-II Ingest | ZFS + ECC Data Integrity | 100% Open-Source OS
Backers: 64
Average Daily Pledges: HK$11,028
Average Pledge Per Backer: HK$7409
Funding: HK$474,195 of HK$9,000
Dates: Nov 20th -> Jan 4th (45 days)
Project By: Harbor Innovations
Backers: 64
Average Daily Pledges: HK$11,028
Average Pledge Per Backer: HK$7409
Funding: HK$474,195 of HK$9,000
Dates: Nov 20th -> Jan 4th (45 days)
Project By: Harbor Innovations
(loading your very own snazzy kicktraq chart)
(loading your very own snazzy kicktraq chart)
(loading your very own snazzy kicktraq chart)
(loading your very own snazzy kicktraq chart)
(loading your very own snazzy kicktraq chart)
Now you can share!
Help your backers reach your daily goals and help others see how you're doing. It's easy, with Kicktraq Mini.Here are a few quick snippits of code you can use to embed your personalized widget on your website, in your blog, and even on your favorite forum.
HTML for websites & blogs
Copy HTMLBBCode for forums
Copy BBCodeReddit code for... well, Reddit
Copy Reddit CodeSpecial Note: Kickstarter recently disabled remote embedding for images on project pages, so embedding code doesn't work on project pages (yet) - just everywhere else.
You can read more about it here.
Latest News
PCIe Explained: Expanding Your Nexus AI Station 
December 22nd - via: kickstarter.com
Just popping in to share a fun little update! 🎬 We’ve just released a new video all about PCIe—yep, that handy little interface that makes expanding your setup possible. We took a casual stroll through the generations of PCIe, from Gen 1 all... (Read More)
New GPU Installed: Local LLM Testing on Nexus AI Station
December 19th - via: kickstarter.com
Quick update from the lab 👀 We’ve just received an AMD Radeon R9700 32GB AI inference GPU, and the first thing we did was install it straight into our Nexus AI Station to start real local LLM testing. After opening up the unit and installing the... (Read More)
Nexus AI Station at AMD Embedded Summit
December 18th - via: kickstarter.com
We’re excited to share that Nexus AI Station was invited to attend the AMD Embedded Summit on December 16–17, 2025. At the event, we showcased Nexus AI Station alongside industry partners and had the opportunity to connect with developers,... (Read More)
Exciting Update: YouTube Review & Expanded KOL Outreach
December 16th - via: kickstarter.com
Hello everyone, Â We have some exciting news to share with our community this week! Â We were thrilled to see a detailed and objective review of our Nexus device by the well-known creator @leepsvideo on the YouTube His video offers a clear and... (Read More)
HarborOS Community Weekly Development Update
December 10th - via: kickstarter.com
Here’s the latest update on the collaborative development of HarborOS. We’ve made steady progress this week, thanks to the active involvement and feedback from our community. Below is a summary of our current status and ongoing efforts:... (Read More)
Deep Dive: Why we built the Nexus
November 29th - via: kickstarter.com
Hi everyone, We know that backing a hardware project requires trust. You want to know who is actually building this machine and why it exists in the first place. We sat down for a candid, unscripted interview with our founder, Bean, to discuss the... (Read More)
Nexus vs. Mac Studio (The "Apple Tax" Breakdown)
November 27th - via: kickstarter.com
Hi Backers (and Watchers), Since we launched, the #1 question we’ve received is: “Why didn’t you use an Intel chip?” or “Why shouldn’t I just buy a Mac Studio for local AI?” Today, we are going deep into the engineering decision behind... (Read More)
📢 Update: Free Shipping for Super Early Bird & Super VIP
November 22nd - via: kickstarter.com
We’ve made an important update to make the Nexus even more accessible for our earliest supporters. Starting today, Super Early Bird and Super VIP backers will receive free worldwide shipping. If you’ve already pledged under one of these two... (Read More)
🚨 Performance Update from the Lab 🚨
November 20th - via: kickstarter.com
We just wrapped up fresh testing at Emdoor Labs with an RTX 4090, and the results are insane. 📝 Text Generation Performance (Qwen3 32B) 136 tokens → 3.52s → 38.64 tokens/sec 500 tokens → 12.18s → 41.04 tokens/sec 1,893 tokens → 46.74s... (Read More)
If you have additional news that doesn't appear here, shoot us a link!