Washington’s national security machinery is being asked to do more than at any point in a decade, with an active war in the Gulf, a Russia policy pulling in two directions and a Taiwan question that won’t sit still. It’s being asked to do it with fewer people, fewer allies sharing in full, and less money from Congress. CrossGov.com, a site that covers US government with a federal focus, has been documenting that gap piece by piece.

I go to conferences. I keep going. I have booked flights for them, expensed hotels for them, and I will do it again this autumn, so nothing that follows is an argument for staying home. What I've stopped doing is pretending the sessions are why I went. The keynote is theatre. Not a bad keynote, not an unusually promotional one, but the format itself. A person walks on stage and, over forty-five minutes, delivers information you could have absorbed in four minutes of reading.

I looked at the public repository at github.com/hypercrux/hypercrux and the project site at hypercrux.com, and examined the published architecture, API, benchmarks, testing methodology and project structure. This is a preliminary engineering assessment based on the repository documentation, not yet an independently executed source-code audit. It covers the released 0.x line; a second engine in development, the Beta, comes up further down. My overall impression is favorable. HyperCrux has a more coherent product proposition than many experimental multimodel databases.

Friday’s wire, minus the duplicates in eight languages. What’s worth a click, then a short tour of five small infrastructure projects headed to Cloudflare Connect. Press releases worth reading Tesla posts Q3 2026 numbers. Production, deliveries and energy deployments for the third quarter are out, the usual early read before earnings. Thales on frontier AI as an attack tool. Thales has put out a global framework for protecting critical organizations against attacks built with frontier AI models.

2G Energy, the German maker of gas engine power plants, has received a 275 MW order from Energy Vault for AI and data center projects in the US, including hyperscaler sites. Deliveries run from the fourth quarter of 2027 through the third quarter of 2028. The value wasn’t disclosed. Power that doesn’t wait for the grid The reason for the deal is simple. Data centers need power now, and grid connections in many US regions take years.

CGI, the Canadian IT services giant, has formed a strategic partnership with D-Wave Quantum to move enterprise clients from quantum experiments to production. It builds on more than a year of joint work. Under the deal, CGI adds D-Wave’s quantum platforms to its advanced technology services. It will help clients find use cases worth pursuing, test them, and then deploy them. The two will also co-develop quantum-enabled solutions. Where quantum is earning its keep The focus is optimization, the kind of problem where the number of possible answers explodes, like schedules, routes and allocations.

Cognex has launched the In-Sight 1750 Series, a new AI-powered reader for wafer and panel IDs. It’s aimed at a problem most people never think about but every fab deals with daily: reading the tiny serial marks laser-etched onto each wafer. Those IDs are how a fab tracks every wafer through hundreds of process steps, and later through packaging. When a reader can’t make out a mark, the tool stops and waits for an operator.

M31 Technology and Ambiq have co-developed foundation IP for TSMC’s N12e process that cuts standard-cell leakage power by about 50%. It’s the kind of gain that decides whether a wearable lasts a day or a week. N12e is TSMC’s 12nm platform for low-power and IoT chips. It’s a mature node where cost and battery life matter more than raw speed. Ambiq builds ultra-low-power chips for always-on devices like smartwatches, hearing aids and sensors, and it needed building blocks tuned to its own power targets.

Exchange Traded Concepts and xETFs have launched NECK, an actively managed ETF on NYSE Arca built around one idea: in the AI build-out, whoever controls the scarcest input has the leverage. “The AI buildout can only move as quickly as its scarcest inputs allow,” says Johnny Wu, founder and CEO of xETFs. “Those constraints will not remain fixed.” What it targets NECK is meant to look past the obvious AI names to the pressure points, meaning companies where supply lags demand, pricing power is shifting, or the position is hard to copy.

PATEO, the Hong Kong-listed Chinese automotive supplier, has signed a strategic memorandum of understanding with Arm to work on physical AI for cars. The two signed in Shanghai on September 22. PATEO calls itself a Tier 1 supplier with “software, hardware, chip and cloud” capabilities. Its core business is intelligent cockpits and connected-car systems for Chinese automakers. Arm designs the processor architecture inside most automotive chips. What they’ll work on The MoU covers next-generation AI workloads in the vehicle, meaning systems that perceive and interact rather than just respond.