Den Do

Tlaaaa co bao sa (@)_(@) Building with @YOM_Official

[YOM @YOM_Official]

A centralized data center goes down thousands of workloads stop.

A node in a decentralized network goes offline the network routes around it. No single point of failure.

No single company controlling capacity.
No single line to cut.
Distributed infrastructure isn't just cheaper. It's harder to break.

That is how we built YOM.

https://t.co/EvAydwgc0G

[YOM @YOM_Official]

Your gaming PC has more compute than it uses.

Between sessions, the GPU sits idle. The hardware runs, the electricity bill runs, the compute does nothing.

YOM routes that compute to real workloads gaming and AI inference.

No mining. No wasted energy. Useful work, real demand.

🛠️ https://t.co/EvAydwgc0G

[YOM @YOM_Official]

A GPU that renders a game session at 7pm can process an AI inference job at 3am.

One machine. Two workload types. Zero idle time wasted.

YOM's network routes the right workload to the right node at the right time.

Operators earn from utilization, not speculation.

That's what useful infrastructure looks like.

[WalterGG @walter___gg]

Downloads. Patches. Storage warnings. Update queues.

The wait before the game somehow became part of the game.

[YOM @YOM_Official]

Traditional mining spends energy to secure consensus.

YOM uses compute to render real games for real players.
One protects a ledger.

The other delivers a product people can use instantly.

DePIN is not a new version of mining. It is a different model entirely.

[YOM @YOM_Official]

Strip away the noise.

$YOM isn't built for speculation. It's the settlement layer for decentralized compute.

Publishers stream games → nodes deliver compute → $YOM settles the transaction on-chain.

No fluff. Just utility powering interactive media at scale.
https://t.co/fkSwG4YC2M

[WalterGG @walter___gg]

Every "next-gen" console launch: same promise, same price jump, same excuse.

We've stopped expecting anything different. Maybe that's the problem.

[YOM @YOM_Official]

Every new gamer can make the YOM network stronger.

More local demand means more regional nodes come online.

More nearby nodes means compute gets closer to players.

That is how YOM is designed to scale: not by adding friction, but by improving coverage and latency over time.
https://t.co/EvAydwgc0G

[YOM @YOM_Official]

This week on YOM 🛠️

→ Why the hardware model is breaking and what replaces it

→ How publishers skip app stores with instant play (+1,316% DAU case study)

→ Heading to @developconf Brighton, July 14–17

→ Town Hall Ep. 80 Xbox layoffs and what it means for decentralized gaming

→ MiCA fully enforced $YOM was compliant from day one

→ The network loop: more gamers → more compute → more $YOM settlement

Building continues. 🫡

[WalterGG @walter___gg]

Owning 300 games and being able to play maybe 20 of

them on your actual setup isn't a library. It's a museum.

[YOM @YOM_Official]

The YOM network runs on a simple loop.

More gamers using platform → more compute consumed → more network activity → more $YOM settlement on-chain → more operators joining to serve demand.

Token utility grows when the network grows.

That's the design.

[WalterGG @walter___gg]

Every year the checklist gets longer: better GPU, more storage, faster internet, a newer console.

And every year the actual game stays exactly the same size it always was.

[YOM @YOM_Official]

Publishers integrating YOM can bypass the traditional app-store funnel.

No downloads.
No store install friction.
No 30% platform fee on streamed access.

A player clicks a link. The game streams instantly from the nearest available node.

In the Martians vs Roughnecks case study, replacing a multi-GB download with instant play drove a +1,316% DAU increase.

That is what removing the barrier does.
https://t.co/AlRT6r6mwt

[Den Do @Do_denTh]

@walter___gg gm

[Den Do @Do_denTh]

#YOM @YOM_Official

[Den Do @Do_denTh]

@YOM_Official yom

[masYNYa @KijAkubovs86334]

🚨 A PORTABLE HACKER KIT WITH A RASPBERRY PI 5 (16GB RAM + 512GB NVME) NOW RUNS LOCAL LLMs — AND KILLS $400/MONTH IN AI SUBS 💀

Not just a pentest rig. A pentest rig that also thinks — without ever asking the cloud for permission.

Pause at 0:08. Look at the stack.

A gloved hand holding up a mobile mesh of hardware. On top: a Raspberry Pi 5 with the official green M.2 HAT+ sitting on the GPIO header, an NVMe SSD tucked into it (red label visible). A Panda Wireless 300 Mbps USB adapter — the classic packet-injection dongle — plugged into a side port. A USB-C cable powers everything. An antenna marked "S22" rises from the back. Under the Pi: an Inseago Verizon 5G MiFi router with a touchscreen, "WiFi ON, Total: 2, Menu / Help" visible.

The whole rig sits on a Lamicall stand and fits in a shoulder bag.

Pause at [0:35]. Look at the second layer.

The Inseago 5G router underneath the Pi. Its display: "5G, Data usage may not be available at this time, WiFi Name/Password." That is his internet, off the grid, wherever a Verizon tower reaches. The finger pointing at it is the tour.

The build, layer by layer:

→ Raspberry Pi 5, 16 GB RAM, 512 GB NVMe SSD on the M.2 HAT+
→ Panda Wireless 300 Mbps USB WiFi — packet injection capable
→ Braided high-gain antenna for capture/deauth work
→ Bluetooth analyzer for BLE device audits
→ Inseago Verizon 5G MiFi router — his mobile internet
→ Powered off a single USB-C bank
→ Everything wrapped in a shoulder bag

What it does on the offensive side:

→ Evil-twin WiFi tests (spin up a fake AP, capture credentials)
→ Deauth attacks in controlled environments
→ Bluetooth Low Energy audits
→ Packet captures that never leave the case
→ Remote SSH via the 5G uplink

What it does on the AI side:

→ Ollama installed, pointed at localhost
→ Llama 3.2 3B / Qwen 2.5 7B quantized — usable chat speeds on CPU
→ 14B models fit in RAM but crawl — realistic for background jobs, not for live coding
→ Claude Code environment variable points at the Pi
→ Every token generates on the same board as the packet captures
→ Zero traffic leaves for API endpoints

The stack he cancelled the same day the rig went mobile:

→ ChatGPT Pro — $200/month
→ Claude Code Max — $200/month
→ Combined annual: $4,800
→ New annual: ~$25 in electricity for the Pi + the price of his data plan (which he was paying anyway)

The math nobody in the pentesting space is walking through:

→ Hardware: Pi 5 16 GB ($120) + M.2 HAT+ ($15) + 512 GB NVMe ($40) + Panda adapter ($30) + Inseago 5G router ($200 secondhand) + antennas & cabling ($40) = ~$445 all in
→ Recurring: $2/mo Pi electricity + a data plan he was already paying for
→ Payback vs the subscription stack: about 6 weeks
→ Year-one delta: nearly $4,500

The honest caveat:

→ Pi 5 CPU inference is not going to match Claude Opus 4.8 on a hard reasoning task
→ For live code pair-programming, cloud still wins
→ For summarization, log parsing, ad-hoc automation, quick answers, on-site work with sensitive captures — the Pi is enough
→ The point is not "match the frontier." The point is "cancel the meter."

The old loop is dead. Buy a hacker laptop. Tether to home WiFi. Send every AI query through OpenAI or Anthropic. Watch the API bill quietly climb during a long engagement.

Now the rig is a briefcase. The internet is a MiFi. The model is on the same board as the packet capture. Nothing phones home.

Here's what nobody in the mobile-security space is saying out loud:

The pentester's threat model always included "assume the network is hostile." The AI subscription stack quietly broke that model — every query flew through a corporate endpoint that logs, retains, and analyzes. This Pi puts the model on the same air-gapped board as the capture. That is the actual upgrade.

Most people are still running Kali on a laptop with ChatGPT open in another tab.

Meanwhile one operator is doing the same audit, thinking with a local model, over 5G, in a coffee shop — with a bill of $0 for the AI part.

Bookmark this. Read the article below.

The race for private, mobile AI just left the laptop.

Literally.

[YOM @YOM_Official]

For a decentralized network to work, nodes have to be reliable.

The YOM settlement contract meters every contribution uptime, region, demand served.

Rewards flow based on verified work. Not self-reported numbers.

On-chain. Transparent. Automatic.
https://t.co/fGVltFTpbR

[WalterGG @walter___gg]

What is the single biggest bottleneck keeping you from gaming more right now?