# AI capex — X 热门讨论 (2026-09-20 14:28 UTC)

## @SecScottBessent (Treasury Secretary Scott Bessent) · 09-20 14:05 · ♥615 ↻173 💬40 .@larry_kudlow in @NewYorkSun gets it right. Here are the facts:

The dollar was on one side of 89.2% of global FX trades.

Foreign investors still hold enormous amounts of U.S. assets. The principal stablecoins are dollar-denominated. Treasury buybacks are about adding liquidity and managing the maturity structure, not somehow controlling a $30T+ market.

Median household income in 2025: $87,460 — a record. Official poverty rate: 10.2% — a historic low. Atlanta Fed GDP forecast: 5.1% annualized real GDP growth in Q3.

And that’s before looking under the hood at the private-sector momentum behind capex and the AI buildout.

The latest numbers show continued employment growth, while business investment and capital expenditures have been expanding, as the CapEx comeback story has continued to broaden out, with nearly 20% more equipment investment in Q2 2026 than at the end of Biden’s term.

Americans deserve honest coverage.

In the New York Times’s latest feeble attempt at delivering a hit piece on the greatest economy in history, the discredited journal selectively excludes data that contradicts this dull narrative. Granted, it’s nothing new that they fail to provide their readers with the full story.

Complicated financial realities should not be reduced to easy-to-read sentences that serve a preordained narrative.

This article’s problem is not complexity. It is selectivity.

If capital is supposedly running from America, why does the data keep showing strong foreign demand for U.S. assets?

If investors are “balking” at Treasuries, someone forgot to tell the bidders.

Norway is not looking elsewhere; it is switching from Treasuries to agency bonds. Still US assets.

Netherlands is bringing gold back from North America, including Ottawa as well as New York, reflecting a home country preference.

We see this morning that Saudi Arabia has dropped out of mBridge and others are likely to follow.

And if digital finance is supposed to threaten the dollar, why are stablecoins overwhelmingly dollar-denominated?

The record is not ambiguous. https://t.co/oc2UM7ls6F https://x.com/SecScottBessent/status/2101674041564176625

## @mayflowers2x8 (May) · 09-20 03:14 · ♥62 ↻1 💬65 The camera is not the interesting part. Where the perception work happens is.

Vangrid’s current architecture puts “edge-computed privacy” alongside multi-view ingestion, cryptographic provenance and an enterprise spatial API. That combination points to a deliberate separation of responsibilities inside the data pipeline rather than treating capture as one giant centralized process.

The edge layer handles the first interaction with the physical world. That is important because spatial capture is inherently sensitive: location, surrounding environments and visual information can reveal far more than a conventional dataset field. Moving privacy computation closer to the source gives the system a place to address that problem before the resulting information moves deeper into the network.

Then the pipeline changes character. Multi-view ingestion is about turning separate observations into usable spatial information, while cryptographic provenance gives that output a traceable relationship back to the underlying contribution. The enterprise API sits at the other end, where the resulting spatial data becomes something downstream systems can actually consume.

I read the “zero-capex sensor swarm” concept in the same way. The model is not centered on building a new physical sensor fleet everywhere. It is about coordinating existing edge devices into distributed perception infrastructure and then standardizing what comes out of that network.

That design has an important consequence for DePIN: the scarce resource is no longer simply hardware. It becomes the quality, coverage, privacy handling and provenance of the data generated by that hardware.

This is also why I find the current direction of @vangrid_io more interesting as infrastructure than as a mapping product. The difficult engineering problem is creating a pipeline where physical observations can move from edge devices into machine-usable spatial data without losing privacy context or provenance along the way.

That is the layer Physical AI actually needs: not more images, but a better system for producing reliable spatial information from the physical world. https://x.com/mayflowers2x8/status/2101510162079195491

## @invest_yash (Yash Raj) · 09-20 06:49 · ♥46 ↻7 💬0 Alpha Ideas 2026 — Key Company Highlights

👉MSafe Equipments Ltd • Growth: Targeting ~50% YoY revenue growth • Capacity: 2x capacity expansion underway • Strategy: Rapid scaling of high-margin rental operations

👉Susan Electricals India Ltd • Order Book: ₹150 Cr with ~3-month execution visibility • Strategy: Revenue mix shifting toward HT Cables, MVCC & LT Cables • Capacity: 60% capacity expansion underway

👉Cryogenic OGS Ltd • Positioning: Preferred engineering partner for critical energy infrastructure • Capacity: 4x capacity expansion underway • Expansion: Empanelled with ADNOC & EIL, while diversifying into Railways & Defence

👉Sacheerome Ltd • Revenue Guidance: ₹200 Cr FY27 → ₹250 Cr FY28 → ₹300 Cr FY29 • Margins: Targeting ~25% EBITDA margin • Capex: 4x capacity expansion, supported by a planned ₹60 Cr debt raise

👉 E to E Transportation Infrastructure Ltd • Growth: Targeting 30%+ revenue growth to cross ₹1,000 Cr • EBITDA: Guided at ~12–14% • Key Initiative: Building a proprietary Railway Intelligence Platform

👉L.T. Elevator Ltd • Capacity: 4x capacity expansion underway • Order Book: ₹300 Cr; another ₹200 Cr of new orders expected • Inorganic Growth: Acquired Ricardo and global specialist Dongyang to expand its international footprint

👉Prizor Viztech Ltd • Business Transformation: Moving from assembly-led operations toward fully backward-integrated manufacturing • Tailwind: STQC approval certification expected to support demand

👉Vivid Electromech Ltd • Business: Expanding technology-driven electrical & automation solutions • Execution: Orders worth ~2x FY26 revenue to be executed over the next 9 months

👉Exato Technologies Ltd • Growth Guidance: Targeting 60–70% growth in FY27 • Order Book: Close to ₹600 Cr • Technology: Building AI Infrastructure as a Service (AIaaS)

👉Workmates Core2Cloud Solution Ltd • FY27 Target: ₹210 Cr revenue, ~45% YoY growth • PAT Guidance: ₹24–25 Cr • Visibility: Entered FY27 with ₹159 Cr of assured recurring revenue • Monthly Run Rate: ~₹13.5 Cr

👉Purple United Sales Ltd • Store Expansion: 43 stores (Mar’25) → ~111 stores (Apr’26) • FY27 Outlook: Targeting ~2x expansion to 200+ outlets

Overall Theme Capacity expansion + strong order visibility + new business verticals + technology adoption + recurring revenue models are the common themes across these companies.

Management guidance and conference commentary should be treated as forward-looking statements, not guaranteed outcomes. https://x.com/invest_yash/status/2101564413224603891