La migra and the meta: big tech's growing pains

The Department of Justice has been looking at Andreessen Horowitz for almost a year now — a 112-year-old antitrust statute, the one nobody touches, being used against a venture firm. The issue is a16z has partners on the boards of companies that now compete: Ben Horowitz sits at Databricks, Martin Casado at Fivetran. Nothing exotic on its face. VCs do this all the time. The DOJ is just asking whether it crosses the line.

The companies weren't direct rivals when a16z first invested, but data infrastructure has gotten crowded. What looked like separate lanes has merged. The antitrust question isn't whether these firms compete — they do — but whether a16z's dual board seats give it enough cross-company information to steer markets. The law is old. The application isn't.

Why this matters for us: it shows the feds are actually willing to pull teeth against well-connected VC networks, which means the usual assumption that a16z can't be touched is about to get tested.

Ordos sits in Inner Mongolia — cheap energy, open land, a few hundred kilometers from Beijing. That combination makes it one of the country's hottest spots for data centers, the buildings where the actual computing happens. A Wired investigation maps how this unlikely place has become the backbone of a massive push to build AI infrastructure.

The story isn't about who's funding it or what models are being trained. It's about the physical setup: why this particular patch of northern China, with its grid access and cooling air, keeps attracting the big warehouse-style facilities. Data centers are thirsty for power and space. The geography lines up.

Why this matters for us: the same forces driving compute buildout in one corner of the world are shaping the global supply chain for the chips and energy we'll need — and the prices we'll pay.

Walmart is joining the rest of the world and will accept Apple Pay and Google Pay at select stores starting August 24. The tap-to-pay rollout hits all US stores by the end of 2026 and gas stations by mid-2027.

This has been a long wait. Walmart was one of the last big retailers not to support Apple Pay or Google Pay, and went its own way for years — backing the short-lived CurrentC before launching Walmart Pay in 2016, a QR code app that lived inside the Walmart app. Now the company is finally letting you use whatever wallet is already on your phone.

Why this matters for us: Walmart is where a lot of la gente shop, and having your phone be the wallet instead of a QR code or a card is one less thing to carry and one less app to open.

Walmart is rolling out Apple Pay and Google Pay at its US stores. The company had refused for years, insisting shoppers swipe a card or tap a physical token. Now the NFC readers can read the phone wallets. The rollout is coming in the fall.

The reason is simple: the tap-to-pay readers are already in the walls. Walmart spent big on them years ago and kept the door shut to mobile wallets because it wanted to push its own app and a proprietary tap card. The hardware works fine for Apple and Google — the payments network is the same — but the company had to flip a switch in its backend. It finally did.

Why this matters for us: la migra card is one less thing to lose, and one less app to juggle at the register — a small win for the people who just want to tap and go without standing in line.

The Idaho National Laboratory is taking apart Chinese lidar units to find security holes — a vulnerability scan paid for by companies in the electric and autonomous vehicle space. The work is a government security review, not a product test, and it targets hardware that feeds sensors to self-driving systems and AVs.

Lidar sends laser pulses to map a vehicle's surroundings. When those sensors are made by foreign vendors and shipped into the US supply chain, the question isn't just whether the beams work — it's whether the firmware, the comms, or the firmware update path can be abused. A bad actor could spoof a sensor, feed a false object, or brick the unit remotely. The review is meant to surface those attack surfaces before the parts make it into fleet vehicles.

This is one of a growing set of hardware scans the US government is running on Chinese components used in transportation, communications, and defense. The electric and AV companies funding it are the ones who'd buy the parts, so the review feeds directly into their procurement decisions and compliance checklists.

Why this matters for us: if you're running a fleet or buying sensors for autonomous vehicles, this scan tells you which Chinese lidar units are likely to be cleared for US use — and which ones are sitting in customs because a lab found a back door.

Laptop prices have gone off the rails. Back when $500 bought a decent machine and $900 got you something solid, the difference was build quality and screen — the chips and RAM were cheap. Now RAM has eaten the whole market. Even the $950 Surface Laptop starts with 8GB, which is basically nothing. On the low end, manufacturers are cutting everything: storage, CPU, build quality — all to compete with the $700 MacBook Neo.

The HP OmniBook 3, $520, flips this. It runs on a Qualcomm Snapdragon chip, ships with 16GB of RAM, and has 256GB of storage. It’s not fancy — the build is plastic, the screen is 1080p IPS, no Thunderbolt, no fingerprint reader. But the 16GB means it actually runs Windows 11 without choking, and the Snapdragon gives you all-day battery and always-on LTE for the $10/month SIM card. You can buy the SIM yourself at a bodega or Walmart — not some corporate plan.

For a primos who needs a laptop for school, a side hustle, or just to keep up — this is the one. It’s not a treat. It’s a tool. And at $520 with 16GB, it’s the only cheap laptop that doesn’t make you regret the purchase six months later.

Why this matters for us: when a $520 machine finally ships with 16GB of RAM, the next time the government asks you to file a form online or show up to a job interview via Zoom, you won’t be stuck on a machine that chokes.