On Moonshots EP #284, the panel read out the performance figures OpenAI published for Jalapeño, the inference chip it built with Broadcom, and argued that inference is moving off NVIDIA. One panelist went further, imagining an "OpenAI Compute" cloud that rents capacity to competitors — possibly even hosting an Anthropic model. Dave Blundin explained why CUDA no longer locks buyers in at inference time, and why NVIDIA's next defense is the networking between chips.
On The Diary of a CEO, critic Ed Zitron praises a chatbot for reading a troubleshooting log and for helping fix his son's Minecraft mod, then argues that neither is worth a trillion dollars. The host counters with his fiancée's one-woman business and his chief of staff's inbox. The argument turns on tokens, subscription rate limits and who is paying the real bill.
On Training Data, Parallel Web Systems founder Parag Agrawal traces how agents multiply web searches — from a weekly credit-risk review across 10,000 small businesses to the meeting-prep agents that run hundreds of searches before his own calls — and forecasts a web that, in a couple of years, tells agents when something worth acting on has changed.
On Training Data, Parag Agrawal explains how his company Parallel entered web search without first building a giant index: it launched a search agent that crawled after a request arrived, replaced outsourced human data collection for insurance, sales and finance customers, and treated the index as a latency optimization to be grown later. He describes the agent-specific architecture behind it, the 200-millisecond Turbo mode Parallel announced in July, and a Google Cloud deal that puts Parallel Search beside Google Search as a grounding option.
Asked about allegations that Chinese labs extracted capabilities from Claude, Alvin Graylin argued that access to another model’s answers cannot explain every engineering advance. The Moonshots exchange turned on three distinctions: legitimate distillation versus prohibited extraction, query bills versus development costs, and learning from outputs versus improving the machinery behind them.
Alvin Graylin argues that China is competing to spread useful AI through industry and overseas developer communities, rather than betting everything on reaching general intelligence first. Provincial competition and open-weight models help explain his account, though the policy contrast is not absolute: America’s AI Action Plan also explicitly promotes adoption.
Alvin Graylin argues that AI can become more useful while earning less for the companies financing its infrastructure. His warning centers on cheaper models and local computing weakening cloud revenues, just as NVIDIA proposes financing platforms intended to mobilize more than $500 billion of outside capital.
On Moonshots, Ramez Naam pointed to the brain’s modest power needs and children’s ability to learn from relatively little data. Co-host Alex countered with a rack of chips producing text thousands of times faster than one writer. Their disagreement connects AI’s energy bill to a larger question: how much improvement can more computation buy?
Ramez Naam passed on Panthalassa’s early Bitcoin-mining pitch, then invested twice in 2026 at much higher valuations. The company now proposes wave-powered AI computing, cooled by seawater and connected by satellite. Its $140 million Series B is intended to support an Oregon pilot factory and northern-Pacific pilots; cheap electricity and longer-lived chips remain prospective benefits.
Naam predicts that AI-assisted code conversion could weaken NVIDIA’s software lock-in in 2026–2027, making rival chips easier to use. The panel’s counterargument: fast connections between chips still matter, even when the work shifts from training models to answering users.
BDH-CQ’s authors report solving 118 of 400 public ARC-AGI-1 tasks at an estimated inference cost of $0.00070 per task, with up to two candidate answers. On Moonshots, Emad Mostaque welcomed architectural experimentation; panelist Alex questioned whether this design offered progress beyond a specialized benchmark.
The Moonshots panel described The Cully Hill Boys as a 110-minute AI feature made by 28 people in four weeks for roughly $2 million, half of it spent on computation. Emad Mostaque expects cheaper video generation to cut that bill sharply. But the film’s published production materials are study-only, and cheaper footage does not resolve performers’ rights or the need for creative direction.