Topic
Running trained models to answer requests, together with the chips, schedulers and serving software used to do it cheaply and quickly.
Cerebras shares traded around $165 in early October, below the $185 IPO price, after the company's lockup expired on October 1. At that level, buyers who paid the $350 opening price in May hold less than half of what they spent.
Reality
Goodfire launched agent monitors on Baseten that it says watched about 1,500 Kimi K3 sessions for $51, against $233 for a model reviewing every step. Every figure so far comes from Goodfire's own tests, so operators still have to measure misses and false alarms on their own traffic.
Reality
Broadcom is arranging more than $50 billion in financing for OpenAI's Jalapeno chip, according to Bloomberg. The talks have not settled a borrower or security, so the risk of a chip still in development could sit with OpenAI, Broadcom or the lenders.
Perspective Coverage
4 publishersReality
OpenAI has contracted 750MW of Cerebras wafer-scale inference capacity, delivered in three 250MW segments by the end of 2028. The dates are contractual, but neither company has published pricing, latency figures or a plan for which customers get access.
Reality
LlamaIndex runs a workload that is 75% inference on rented GPUs, parsing millions of document pages a day that arrive in bursts. Its CEO's account at the Fully Connected event describes a buyer for whom guaranteed capacity matters more than owning hardware.
Reality
GMI Cloud raised $668 million, two-thirds of it through a $445 million credit facility led by CTBC and the rest as a $223 million Series B. Repaying debt on that scale depends on the more than $600 million in contracted annual revenue GMI reports and on how quickly it puts GPUs into production to serve it.
Reality
CoreWeave now rents Nvidia's Vera Rubin NVL72 racks, with first customer Cognition reporting up to 4.8x the token throughput on SWE-2 inference. That figure is a best case from a single inference workload, and it is the only customer result behind the launch.
Reality
The independent inference-silicon pitch did not lose an argument. It was licensed, hired away, and reissued as a neocloud that runs Nvidia systems.
Perspective Coverage
4 publishersReality
The inference challenger now sells Nvidia clusters and takes Nvidia money. Buyers who priced in a non-Nvidia alternative should re-read their renewal terms.
Perspective Coverage
5 publishersReality
All seven CoreWeave customers running workloads in a 13-buyer SiliconANGLE study expect to grow their spending or workloads. Three of the ten buyers past evaluation are not running on it, and hyperscalers still hold the rest of the application stack.
Reality
Jalapeno's lead is measured against last generation and the volumes are tiny, but a first-pass ASIC clearing Nvidia, AMD and Google parts reprices the design barrier, not the supply chain.
Perspective Coverage
7 publishersReality
Custom silicon made up 73% of the $16.7bn AI quarter. Hock Tan says that line doubles twice by fiscal 2028. He has already secured wafers and memory through 2028 to serve promises only six buyers have made.
Perspective Coverage
9 publishersReality
Crusoe's Series F values the Denver data centre builder at $30.9bn and pays for capacity customers have already contracted. One gigawatt is running, and the company now makes its own circuit breakers and industrial controls.
Perspective Coverage
4 publishersReality
Hock Tan told CNBC that Broadcom's $115bn and $230bn AI chip targets for fiscal 2027 and 2028 are intact after the stock fell 4.8%, and the company has lifted this year's AI guide to $58bn from $56bn.
Reality
A SiliconAngle opinion piece written from inside the AI semiconductor business says commodity inference would grow the market. Teams that ration tokens today still have to decide whether their caps are configuration or architecture.
Reality
CNBC reports both labs have sounded out powered sites of roughly 20 to 30MW in the UK, the Nordics and the US. A hall that small is usable because inference requests can be routed independently.
Reality
Qualcomm's Tony Pialis told the AI Infra Summit in Santa Clara that tokens per watt is the new key metric in the AI war. The vendors in the room disagreed about which memory architecture gets them there.
Reality
Cisco and Splunk spent the .conf26 keynotes on trust and on who is authorized to act. The two items named as shipping are a token-cost meter available now and a dataset join gated on holding two licenses.
Reality
A $40 million Series A led by Spark Capital funds 8-by-8-foot modules that plug straight into solar panels and run on the midday surplus grids refuse. Rune puts the saving at about 90% per megawatt of capacity.
Reality
Dario Amodei's call for slower model development knocked 4.8% off Broadcom on Monday, and Hock Tan told CNBC his $115 billion and $230 billion AI chip targets for fiscal 2027 and 2028 have not moved. Anthropic is due to become his largest custom chip customer next year.
Reality
Science · September 15, 2026 · One report1 publisher
Leadership · September 15, 2026 · One report1 publisher
Product · September 11, 2026 · Confirmed2 publishers