
SpaceX AI Device - Real Prototype or Investor Theater?
The WSJ says SpaceX showed investors a Grok-powered AI handset before its IPO. Musk called it false. We examine the hardware stack and what the denial actually means.
They summarize our coverage. We write it.
Newsletters like this one rebroadcast our headlines - often without the full review, the source reading, or the analysis underneath. Our weekly briefing sends the work they paraphrase, straight from the desk, before they get to it.
Free, weekly, no spam. One email every Tuesday. Unsubscribe anytime.

The WSJ says SpaceX showed investors a Grok-powered AI handset before its IPO. Musk called it false. We examine the hardware stack and what the denial actually means.

Meituan open-sources LongCat-2.0, a 1.6T MoE model trained on 50,000 Chinese ASICs that secretly topped OpenRouter under the alias Owl Alpha.

Anthropic is in early talks with Samsung to develop its first custom AI chip on a 2nm process, making it the last major frontier lab to enter the custom silicon race.

d-Matrix Corsair is an SRAM-based in-memory compute ASIC in production since June 2026, targeting 10x faster and 5x more power-efficient LLM inference vs GPU baselines.

OpenAI's first custom AI chip, co-designed with Broadcom on TSMC 3nm, targeting 50% lower inference cost than GPU alternatives.

Transformer ASIC startup Etched comes out of stealth with first-pass silicon on TSMC N4P, $800M raised, and more than $1B in signed customer contracts.

South Korea announced an 800 trillion won chip investment plan to build four new fabs with Samsung and SK Hynix in the country's southwest, targeting AI memory dominance through 2035.

Qualcomm confirmed a $3.9B acquisition of AI software startup Modular and is in advanced talks to buy RISC-V chip designer Tenstorrent for up to $10B, betting that open hardware and a portable compiler can dislodge Nvidia from the data center.

Cerebras beat Q1 revenue estimates with 92% growth but a dramatic margin guidance cut sent shares down 20%, erasing nearly all gains since its landmark semiconductor IPO.

OpenAI and Broadcom unveiled Jalapeño, a custom ASIC on TSMC's 3nm node with eight HBM stacks, targeting 50% cheaper inference than current GPU-based alternatives.

Amazon CEO Andy Jassy hints the company will sell Trainium3 racks directly to outside data centers, citing a potential $50B revenue run rate and sold-out chip supply.

Broadcom, Apollo, and Blackstone launched the $35B AI XPV Platform on June 9, targeting 20GW of custom AI silicon through 2028 with Anthropic as the anchor customer.