buildConfirmed2 publishers Liquid AI's open-weight d1-3B answers classification questions in a single forward pass, taking 16 ms on a Jetson AGX Thor and 50 ms on an Orin Nano. That puts routing and intent tagging on edge boards without a large model, on figures Liquid measured itself.
Perspective Coverage
3 publishers
- Builder
- Builder 55%
- Operator
- Operator 30%
- Investor
- Investor 15%
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+20
- Incentives70
- Confidence60
Lola Vision Systems targets the 200 hours of setup between an AI model and a new chip
Lola Vision Systems will license its compiler on existing chips, aiming at what its founder says can be roughly 200 hours of manual setup per new chip. Cheaper porting would let edge teams test a second board, though the company has not said how many of those hours its software saves.
Reality
- Evidence25
- Adoption10
- Hype gap+30
- Incentives80
- Confidence35
SiMa.ai raised $150 million at a $1.45 billion valuation to scale its development environment and build its next embedded chip. For device makers on Nvidia's Jetson boards, the case turns on whether that software ports a model in the days or hours SiMa.ai claims.
Perspective Coverage
4 publishers
- Builder
- Builder 33%
- Operator
- Operator 20%
- Investor
- Investor 47%
Reality
- Evidence58
- Adoption30
- Hype gap+40
- Incentives65
- Confidence60
Robot makers buy the inference before they sell the robot
Where a robot's inference runs decides what the machine costs to build. SemiAnalysis puts the industry's upfront compute bill in the trillions at a billion robots, paid by manufacturers on every unit before it does any work.
Reality
- Evidence52
- Adoption30
- Hype gap+18
- Incentives55
- Confidence45
buildOne report1 publisher NVIDIA's edge-agent case rests on compact open models matching data-center capabilities
Nemotron 3.5 Lightning touches a tenth of its weights per token, which is what makes an agent loop plausible next to the sensors, but memory still has to hold all thirty billion and NVIDIA's own loop still escalates off-device.
Reality
- Evidence42
- Adoption
- Not scored
- Hype gap+38
- Incentives84
- Confidence54
Amazon triples its Nvidia order, and the 2027-28 GPU queue closes early
Two million Blackwell Ultra and Rubin parts, booked five months after a one million commitment. Custom silicon did not substitute for Nvidia; it added to the bill.
Reality
- Evidence58
- Adoption62
- Hype gap+22
- Incentives78
- Confidence60
Generalist raises a reported $200M a week after Gen-1.5, and the automation bill moves line items
The model teaches robot arms from hand demonstrations instead of per-task code. Generalist's own figures put completion at 83% with a few examples, which still leaves a person at the station.
Reality
- Evidence30
- Adoption12
- Hype gap+38
- Incentives72
- Confidence40
Instacart's real bet is not delivery. It is owning the grocery aisle's missing data
The Caper Cart's cameras, scale and edge compute turn a store trip into a measurable channel. The scale claims stay vague, but the ownership question does not.
Reality
- Evidence30
- Adoption38
- Hype gap+42
- Incentives78
- Confidence42