E67: 12/6/24
Amazon's "Nova" model family, Microsoft Copilot Vision, OpenAI's o1 in the wild, a $10B Meta datacenter, more crazy humanoid robots, etc.
Welcome back after a week off for Thanksgiving! Its a crazy one with the “12 days of OpenAI”, AWS Re:Invent, and more. Enjoy!
PS- let me know if you’re in Vancouver this Monday/Tuesday for the NeurIPS Conference… would be great to meet up!!
PNW AI News:
At Re:invent, Amazon announce its own internal set of AI foundation models dubbed "Nova", hosted in AWS Bedrock, including:
Nova Micro, a text-only model with low latency
Nova Lite, a low-cost multimodal model for image processing
Nova Pro, a multimodal model that combines accuracy, speed and cost
Nova Premier, its most expensive model that is capable of complex reasoning like OpenAI's "o1"
Nova Canva, for image generation
Nova Reel, for video generation
They're also adding new AI models from Luma AI and Poolside to the core Amazon Bedrock genAI managed service, not holding exclusivity to its own 'Nova' LLMs
Amazon also announced a series of updates to Amazon Q Developer, its AI assistant for software development, including the ability to use AI to migrate companies off of legacy Microsoft .NET applications to Linux
AWS CEO Matt Garman called this the "easy button" to get off Microsoft
Microsoft releases preview of voice-enabled ‘Copilot Vision’ AI browsing companion for its Edge browser that 'sees' the browser as a user would
Amazon and Anthropic unveiled plans for Project Rainer, a massive AI supercomputer powered by >100K Trainium2 chips (5x larger than Anthropic’s current top model) that promises to be the world's largest AI system
Key AI Product/Research Updates:
OpenAI just released its o1 model out of preview during the first day of its ‘12 days of OpenAI’ event, alongside a new $200/mo ChatGPT Pro subscription tier
During this announcement, they detail o1's ability to generate a design for a 1GW datacenter in space. Sound familiar?? Lumen Orbit
Google introduced the PaliGemma 2 model family, designed to analyze images and detect emotions, offering contextually relevant captions and descriptions of actions and emotions
Has sparked controversy, as it requires fine-tuning for accurate emotion recognition on an "subjective dataset"
Clone Robotics introduced Clone Alpha, an (extremely) humanoid robot featuring synthetic organs and water-powered artificial muscles, with 279 officially available for preorder in 2025
Google DeepMind’s Genie 2 Turns Single Images into Playable 3D Worlds, dubbed a foundation world model (LWM)
DeepMind just unveiled GenCast, an AI weather forecasting system that surpasses the accuracy of the world's leading forecasting model, producing reliable predictions for 15-day forecasts in minutes rather than hours
Tencent just released HunyuanVideo, a new open-source, open-weights, 13B parameter AI video generation model that beats top closed rivals in testing
‘Godmother of AI’ Fei-Fei Li’s startup World Labs just revealed its first major project, an AI system that can transform any image into an explorable, interactive 3D environment that users can navigate in real-time through a browser
ElevenLabs unveiled Conversational AI, which allows users to seamlessly add voice capabilities in 31 languages to AI agents with features like ultra-low latency, LLM flexibility, advanced turn-taking, and more
Key AI Business/Investment Updates:
Trump names David Sacks as White House AI and Crypto czar, now tasked with steering federal AI policy, and will also chair the Presidential Council of Advisors for Science and Technology
Expect Sacks to coordinate closely with other high-profile tech leaders on responsible deregulation of AI accelerationism
xAI, Elon Musk’s AI company, has raised $6B from undisclosed investors, according to a filing with the SEC on Thursday
OpenAI is reportedly exploring adding advertising into its AI products as it seeks new revenue streams, with CFO Sarah Friar confirming the company is evaluating an ads model
Meta is putting a $10B AI data center in Richland Parish, Louisiana with an expected open date in 2030
In an NYT interview, Sam Altman provided new numbers on ChatGPT’s adoption, including 300M weekly active users, 1B daily messages, and 1.3M U.S. developers on the platform
Anduril announced a partnership with OpenAI to develop AI-powered aerial defense systems to protect U.S. and allied forces from drone threats
PNW AI/ML/CV Fundings:
Spexi (Vancouver, BC) raised $11.5M in Series A funding led by Blockchange Ventures, joined by Protocol Labs and Dapper Labs
Professional drone imaging services intended to help pilots fly projects in an area and earn money, using CV to provide insights from that imaging – TFTD $17M
CEO: Bill Lakeland
RL Core (Edmonton, AB) raised $5M of Seed funding led by TQ Ventures, joined by Flying Fish Partners
AI system intended to improve the efficiency and performance of industrial control systems – TFTD $5.79M
CEO: Martha White
Sight Glass (Portland, OR) raised $3m in Seed funding led by Base10 Partners
AI-powered investor relations management software designed for investment firms and asset owners – TFTD $3M
CEO: Thomas Buley
OraQ AI (Calgary, AB) raised $2.6M in Seed funding from Angel MD and Startup TNT
Developer of a clinical decision support system designed to empower patients through improved treatment strategies, advanced assessments, and detailed risk profiling – TFTD $3.6M
CEO: Amreesh Khanna
Lightstrike (Calgary, AB) raised $800K in Pre-seed funding led by Sprout Fund, joined by the Binnacle Group, Weave VC, Daglar Cizmeci and other undisclosed investors
Software dependency management application intended to provide control over tech ecosystems using AI – TFTD $800K
CEO: Vernon Briggs
Kinesis (Redmond, WA) raised an undisclosed Pre-seed round from Brilliant Minds and other investors
Gathers idle GPU and CPU power from around the globe, transforming it into a unified compute cloud tailored to your needs – TFTD Unknown
CEO: Baris Saydag