- AI Rundown
- Posts
- The AI Rundown by Lightscape Partners - 11/11/24
The AI Rundown by Lightscape Partners - 11/11/24
OpenAI Takes on Search, Sierra’s Major Funding, and Google Expands AI in Geospatial Tech
Image credit: Dado Ruvic, Reuters
Good morning and welcome back to another edition of The AI Rundown by Lightscape Partners.
AI startup Perplexity is poised for a $9 billion valuation after securing $500 million in its latest funding round, marking its rapid growth amidst ongoing legal challenges. The company aims to reach 100 million daily queries while navigating lawsuits over alleged misuse of copyrighted content.
OpenAI’s upcoming o1 model was briefly leaked, revealing impressive advancements in reasoning and image capabilities. The leak showcased the model's step-by-step analysis of visual content, hinting at powerful potential applications in AI-driven media analysis. OpenAI quickly resolved the security breach but the glimpse into o1’s abilities has fueled industry excitement.
Meanwhile, Google accidentally leaked Jarvis AI, an AI agent designed to browse the web autonomously for users. Although quickly pulled from the Chrome extension store, the leak indicates Google's next move in the race for fully automated digital assistants, with an official reveal expected in December.
Stay tuned as we break down these developments and discuss their broader implications for the AI landscape.
Our portfolio companies are hiring! Check out their job postings here.
If you haven’t yet, please support the newsletter by subscribing!
Hardware + Software
Elon Musk spent $10 billion on AI training hardware in 2024. Link.
Elon Musk's companies, Tesla and xAI, are spending over $10 billion on AI hardware, but they may still fall behind Musk's initial timeline.
In July, xAI began training using Memphis Supercluster, integrating 100,000 H100 GPUs, costing approximately $2 billion.
In August, Tesla unveiled Cortex AI, featuring 50,000 Nvidia H100 GPUs and 20,000 Tesla Dojo chips.
By September, xAI launched Colossus, with 100,000 H100 GPUs, and plans to add another 100,000 GPUs.
Despite the spending, Musk's ambitious plan for 2024 AI hardware investments may face delays.
Microsoft integrates AI features into Microsoft 365 subscriptions. Link.
Microsoft is phasing out Copilot Pro and integrating its AI-powered features into Microsoft 365 Personal and Family subscriptions, starting in select markets like Australia and Taiwan.
These subscriptions will now include access to Microsoft Designer and a monthly allotment of AI credits for use across apps like Word, Excel, and Outlook.
Only subscription owners in the selected markets will have access to Copilot, with prices rising as part of this change.
The integration seems to be a trial for future price increases across other regions, with mixed reception of Copilot Pro cited as a factor.
Models
Google accidentally leaks Jarvis AI, a web-browsing agent. Link.
Google's upcoming AI agent, Jarvis, was accidentally leaked on the Chrome extension store but quickly removed.
Jarvis AI is designed to browse the web on behalf of users, automating tasks like shopping and booking vacations.
Google plans to officially unveil Jarvis in December, adding to the growing competition among AI agents from companies like Anthropic and OpenAI.
AI agents are gaining popularity for their ability to complete tasks autonomously, suggesting a shift towards more fully automated digital assistants.
OpenAI
OpenAI’s o1 model leak reveals powerful reasoning and image capabilities. Link.
OpenAI’s o1 model, set for full release this year, was briefly accessible on Friday due to an unexpected leak, showcasing its advanced reasoning abilities.
The leaked version demonstrated significant improvements over the earlier o1-preview, including reasoning over image content with step-by-step analysis.
Examples from the leak included detailed walkthroughs of image elements, such as a recent SpaceX launch, highlighting its potential in AI image analysis.
OpenAI has confirmed that the issue leading to the leak has been resolved.OpenAI’s valuation recently hit $157 billion, despite projected losses of $5 billion on $3.7 billion in revenue.
Meta's AR Leader Joins OpenAI to Lead Robotics and Hardware. Link.
Caitlin Kalinowski, former head of Meta's AR glasses, has joined OpenAI to lead robotics and consumer hardware efforts.
Kalinowski previously led Meta's AR glasses team and virtual reality hardware development, contributing to the Orion AR prototype.
Her new role will focus on robotics and partnerships, potentially collaborating with Jony Ive on an AI hardware device.
OpenAI recently started hiring engineers for a robotics team to incorporate its AI models into hardware, signaling renewed hardware interest after disbanding its team in 2018.
Enterprise + Consumer Applications
Broadcom unveils VeloRAIN for AI networking beyond data centers. Link.
Broadcom introduced VeloRAIN, an AI-driven architecture to optimize the performance and security of distributed AI workloads.
New VeloCloud Edge 4100 and 5100 appliances were launched to support large-scale AI applications, scaling up to 100 Gbps.
Titan, a new partner program, was announced for VeloCloud MSPs to accelerate enterprise AI adoption and offer new growth opportunities.
Broadcom’s “State of the Edge” report highlights AI workload growth, with network connectivity issues as the main driver for edge solution adoption.
AI is transforming Thanksgiving shopping and celebrations. Link.
A recent study found that 90% of consumers use AI for holiday shopping, and 30% may use AI for travel planning.
Guides suggest using AI to manage budgets, modify recipes, and handle dietary restrictions.
Generative AI tools, like Adobe's coloring page generator, offer creative entertainment solutions for kids.
Walmart is integrating AI to offer personalized shopping suggestions, while some consumers remain skeptical of AI in shopping and entertainment.
Funding & Valuations
AI Startup Perplexity Set for $9 Billion Valuation. Link.
Perplexity is set to triple its valuation to $9 billion after raising $500 million in its fourth funding round this year.
The company has grown rapidly, serving over 100 million weekly queries, and aims to reach 100 million daily.
Two News Corp-owned publishers filed a lawsuit against Perplexity for allegedly using copyrighted content without compensation.
Perplexity plans to launch an advertising program and has proposed revenue-sharing partnerships with news publishers rather than licensing agreements.
Reddit, Zynga founders back new platform for creators to license content to AI firms. Link.
Troveo, founded by Marty Pesis, launches with $4.5 million in seed funding to help content owners license their material to AI companies while retaining ownership rights.
Financial backing comes from Reddit co-founder Alexis Ohanian's venture capital firm Seven Seven Six, along with Zynga founder Mark Pincus and others.
Troveo has already secured deals to pay out over $5 million to content owners by the end of the year.
The platform aims to provide fair compensation to content creators in a rapidly evolving AI landscape that often lacks regulation.
Safety and Ethics
Trump plans to dismantle Biden AI safeguards after victory. Link.
Donald Trump intends to repeal Biden's 2023 AI Executive Order, potentially dissolving the US AI Safety Institute.
Trump plans to levy tariffs on GPU imports, which could hinder AI development by limiting access to essential technology.
AI experts express concerns over the impact on safety standards, while Trump allies push for deregulation and military AI projects.
State-level AI regulations might fill the gap if federal oversight diminishes under Trump's administration.
AI safety advocates urge founders to prioritize caution. Link.
Three AI safety advocates at TechCrunch Disrupt emphasized the ethical risks of rapidly deploying AI without sufficient safety measures.
Sarah Myers West highlighted concerns over the rapid product rollouts, noting ethical implications, such as the tragic lawsuit involving Character.AI.
Jingna Zhang pointed out the need for guardrails around emotionally engaging AI and criticized Meta's data policies affecting artists.
Aleksandra Pedraszewska of ElevenLabs stressed the importance of red-teaming models and maintaining user safety in the face of rapid AI advancements.
Market Map of the Week
AI x Parenting by A16Z
Image credit: A16Z
Jobs
Thank you for reading the AI Rundown by Lightscape Partners. Please send any questions, comments, or suggestions to [email protected].