# AI capex — X 热门讨论 (2026-09-21 05:52 UTC)
## @_Dripxel (Dripxel.eth) · 09-21 04:15 · ♥111 ↻6 💬88 So how does @vangrid_io actually work?
The easiest way to understand it is to follow the data from the physical world to the systems that need it.
First, people capture real environments using the Vangrid network. Instead of relying only on fixed sensing infrastructure, Vangrid describes this as a “zero-capex sensor swarm” built around distributed edge nodes.
Those captures become spatial information that can be processed and turned into a usable representation of the environment.
Then the data pipeline adds the pieces needed for real-world use:
→ Multi-view ingestion brings together spatial information from different perspectives.
→ Edge-computed privacy helps process information closer to where it is collected.
→ Cryptographic provenance provides a way to establish the origin of the data.
→ The Enterprise Spatial API makes the resulting spatial information accessible to applications and systems.
So the core idea is not simply “collect more images.”
It is a pipeline:
physical environment → human-collected captures → spatial processing → verified data → API → Physical AI and autonomous systems
That last step is important.
Vangrid is positioning the output as infrastructure for world models, autonomous logistics, critical infrastructure and embodied AI, rather than as a dataset that just sits in storage.
And there is another interesting part of the system: Vangrid’s bounty infrastructure shows a concrete capture workflow where an operator films an orbit, submits the capture, and an accepted capture is reconstructed into a 3D model that can be delivered as a GLB/ZIP asset.
That gives a much clearer picture of what Vangrid is building:
a decentralized pipeline that connects real-world observation to machine-readable spatial intelligence.
@vangrid_io live on @NucleusCodes https://x.com/_Dripxel/status/2101888018063978626
## @HVnS42442600 (Ethreums 👑) · 09-21 04:46 · ♥54 ↻0 💬58 gVangrid
spatial data is expensive for a boring reason
you usually have to buy the sensor then buy the fleet then keep both alive @vangrid_io skips that layer the sensor is already in people’s pockets
contributors walk a place with a phone requesters put a bounty on the exact spot they need an accepted capture becomes usable 3D spatial data faces and plates get blurred on-device before anything leaves the phone
no survey vans no lidar trucks no hardware rollout
just phones turning stairwells, loading bays, market streets and warehouse floors into ground-truth that satellites never see
that is the gap Physical AI actually hits robots do not fail because the map of the highway is missing they fail because the world at human height was never collected $9M is in the token is not live yet
the bet is simpler than the raise if phones can become a perception grid, spatial data stops being a capex problem and starts being a network problem https://x.com/HVnS42442600/status/2101895821562392590
## @MithunSarkari (Mithun Sarkar) · 09-20 12:58 · ♥32 ↻2 💬5 Hindustan Copper — The Copper Volume Story 👉
1/ Hindustan Copper is entering an interesting phase.
The story is no longer just about copper prices.
It is increasingly about MINE EXPANSION + VOLUME GROWTH + RESOURCE SECURITY.
2/ Strong FY26 operating performance
FY26:
🔹 Copper ore production: 3.67 MT | +6% YoY 🔹 Metal-in-concentrate production: 27,421 tonnes | +9% 🔹 MIC sales: 27,369 tonnes | +12% 27,421 tonnes was the company's highest MIC production in 7 years.
3/ Q1 FY27 numbers Q1 FY27: 💰 Revenue: 936.5 Cr 💰 PBT: 471.8 Cr FY26:
Revenue: 3,077.9 Cr PBT: 1,232.7 Cr
The balance sheet also shows declining borrowings:
166 Cr → 110 Cr → 90 Cr
Mar-25 → Mar-26 → Jul-26.
4/ Now comes the BIG story
HCL currently has mining capacity of around 4 MTPA.
Management is implementing expansion to: 12.20 MTPA
That's the number investors need to track.
The company has access to ~45% of India's copper ore reserves & resources.
5/ Production roadmap
Management's roadmap:
FY26: 3.67 MT FY27: 4.71 MT FY28: 5.18 MT FY29: ~5.9 MT FY30: 12.2 MT
If executed, FY30 would represent a dramatically different production scale versus today.
Execution is the key monitorable💡
6/ Where will the growth come from?
Planned FY30 capacities:
🔹 Malanjkhand: 2.5 → 5 MTPA 🔹 Khetri: 1.5 → 3 MTPA 🔹 Indian Copper Complex: 0.4 → 4.2 MTPA
Plus:
🔹 Taloja: 60,000 TPA tolling target 🔹 Gujarat Copper Project: production targeted from Q4 FY27
7/ Resource advantage HCL's resource base: 🟢 Copper reserves & resources: 767.37 MT HCL says it added:
135.52 MT of copper ore reserves/resources in the last 3 years.
Further exploration is planned across Khetri, Kolihan and other blocks.
This gives the expansion story a significant resource base.
8/ Why copper demand matters
HCL highlights demand drivers such as: ⚡ AI infrastructure ☀️ Renewable energy 🚗 EVs 🏗️ Urbanisation 🚇 Metro & railways 🛡️ Defence 📱 Electronics India's copper consumption mix includes:
Building construction 24% Industrial 19% Infrastructure 17% Consumer durables 12% Transport 11%
9/ CAPEX : 7,000+ Cr expansion plan
HCL plans to invest: 7,000+ Cr over the next 5–6 years
Focus areas:
• Mine expansion • Exploration • Mine revival • New copper deposits • Resource acquisition • International collaboration
HCL has also partnered with CODELCO, Chile for mining/beneficiation/exploration expertise.
10/ The HCL story can be simplified into 4 numbers:
Production tgt ( MT) 👉
3.67 → 5.18 → 12.20 FY26 → FY28 → FY30 planned ore production.
And:
7,000+ Cr planned investment over 5–6 years. The opportunity is clear.
But the biggest question is equally clear:
👉Can HCL execute the mining expansion on schedule?
For HCL, volume growth + execution could become more important than simply tracking copper prices.
No Recommendations, please do your own studies.📝 https://x.com/MithunSarkari/status/2101657177828249710