OpenAI now projects $125B in revenue in 2029, with $25B of that from new products not yet announced
The Information reports OpenAI forecasts revenue reaching $125 billion in 2029 and $174 billion in 2030, mainly from AI agents, subscriptions, monetizing free users, and potentially affiliate fees.
According to internal documents seen by The Information, OpenAI expects revenue in 2029 to include $29 billion from AI agents, $50 billion from ChatGPT subscriptions, $22 billion from API access, and $25 billion from monetizing free users and other new, unspecified products.
CEO Sam Altman mentioned recently affiliate fees or taking a percentage of sales generated through user searches as possible revenue sources, while CFO Sarah Friar told the Financial Times there are โno active plansโ for selling traditional advertising.
If they hit it, the current valuation ($300B) will be a steal; Google does ~$400B in revenue and is worth $2T.
The Information reports OpenAI forecasts revenue reaching $125 billion in 2029 and $174 billion in 2030, mainly from AI agents, subscriptions, monetizing free users, and potentially affiliate fees.
According to internal documents seen by The Information, OpenAI expects revenue in 2029 to include $29 billion from AI agents, $50 billion from ChatGPT subscriptions, $22 billion from API access, and $25 billion from monetizing free users and other new, unspecified products.
CEO Sam Altman mentioned recently affiliate fees or taking a percentage of sales generated through user searches as possible revenue sources, while CFO Sarah Friar told the Financial Times there are โno active plansโ for selling traditional advertising.
If they hit it, the current valuation ($300B) will be a steal; Google does ~$400B in revenue and is worth $2T.
๐3โค2๐2
Trends in AI Supercomputers
Epoch AI dropped a map of the worldโs 500+ AI supercomputers.
โข Performance doubling every 9 months.
โข Hardware cost & power use doubling every year.
โข xAIโs Colossus already gulps 300 MW โ equal to 250 k homes โ and thatโs only 2025.
If the trendline holds, the 2030 front-runner will burn 9 GW, pack 2 M chips, and sport a $200 B price tag.
The U.S. owns 75 % of todayโs compute muscle while China trails at 15 %.
Epoch AI dropped a map of the worldโs 500+ AI supercomputers.
โข Performance doubling every 9 months.
โข Hardware cost & power use doubling every year.
โข xAIโs Colossus already gulps 300 MW โ equal to 250 k homes โ and thatโs only 2025.
If the trendline holds, the 2030 front-runner will burn 9 GW, pack 2 M chips, and sport a $200 B price tag.
The U.S. owns 75 % of todayโs compute muscle while China trails at 15 %.
๐ฅ7โค3๐2
MIT researchers have developed a "periodic table" for machine learning โ a groundbreaking framework that maps the connections between 20+ classical ML algorithms.
By revealing how these methods relate and overlap, the table opens up new ways for scientists to hybridize techniques, improving existing models or even inventing entirely new ones.
As proof of concept, the team fused two distinct algorithms using this framework and created a novel image classification method โ outperforming current state of the art models by 8%.
By revealing how these methods relate and overlap, the table opens up new ways for scientists to hybridize techniques, improving existing models or even inventing entirely new ones.
As proof of concept, the team fused two distinct algorithms using this framework and created a novel image classification method โ outperforming current state of the art models by 8%.
Tech Xplore
'Periodic table of machine learning' framework unifies AI models to accelerate innovation
MIT researchers have created a periodic table that shows how more than 20 classical machine-learning algorithms are connected. The new framework sheds light on how scientists could fuse strategies from ...
๐ฅ5๐5โค3๐1
Itโs announcement about the new lightweight ChatGPT deep research model (powered by a version of o4-mini) and updated limits confusing
In the end, it's a combination of tasks using "standard" deep research plus additional tasks using the lightweight version:
- Free - 5 tasks/month using the lightweight version
- Plus & Team - 10 tasks/month, plus an additional 15 tasks/month using the lightweight version
- Pro - 125 tasks/month, plus an additional 125 tasks/month using the lightweight version
- Enterprise - 10 tasks/month
In the end, it's a combination of tasks using "standard" deep research plus additional tasks using the lightweight version:
- Free - 5 tasks/month using the lightweight version
- Plus & Team - 10 tasks/month, plus an additional 15 tasks/month using the lightweight version
- Pro - 125 tasks/month, plus an additional 125 tasks/month using the lightweight version
- Enterprise - 10 tasks/month
OpenAI
Introducing deep research
An agent that uses reasoning to synthesize large amounts of online information and complete multi-step research tasks for you. Available to Pro users today, Plus and Team next.
๐ฅ3โค2๐2
Liquid AI introduced architecture called Hyena Edge, a convolution-based multi-hybrid model that not only matches but outperforms strong Transformer-based baselines in computational efficiency and model quality on edge hardware, benchmarked on the Samsung S24 Ultra smartphone.
To design Hyena Edge, researchers used end-to-end automated model design framework โSTAR.
To design Hyena Edge, researchers used end-to-end automated model design framework โSTAR.
www.liquid.ai
Convolutional Multi-Hybrids for Edge Devices | Liquid AI
Today, we introduce a Liquid architecture called Hyena Edge, a convolution-based multi-hybrid model that not only matches but outperforms strong Transformer-based baselines in computational efficiency and model quality on edge hardware, benchmarked on theโฆ
โค3๐3๐2
Why Do Multi-Agent LLM Systems Fail? Despite the growing excitement around Multi-Agent Systems (MAS), they often struggle to outperform single-agent approaches.
Berkeleyโs researchers analyzed 7 popular MAS frameworks across 200+ tasks, identifying 14 failure modes that hinder their effectiveness.
Paper.
Code.
Berkeleyโs researchers analyzed 7 popular MAS frameworks across 200+ tasks, identifying 14 failure modes that hinder their effectiveness.
Paper.
Code.
Google
MAST
In a Nutshell
Despite the increasing adoption of Multi-Agent Systems (MAS) , their performance gains often remain minimal compared to single-agent frameworks. Why do MAS fail?
We have conducted a systematic evaluation of MASs execution traces using Groundedโฆ
Despite the increasing adoption of Multi-Agent Systems (MAS) , their performance gains often remain minimal compared to single-agent frameworks. Why do MAS fail?
We have conducted a systematic evaluation of MASs execution traces using Groundedโฆ
โค5๐3๐2
Stripe is building a NEW stablecoin product, powered by Bridge
If your company is:
- Based outside of the US, EU, or UK
- Interested in dollar access
Send a quick note about your company to stablecoins@stripe.com
If your company is:
- Based outside of the US, EU, or UK
- Interested in dollar access
Send a quick note about your company to stablecoins@stripe.com
๐ฅ5๐ฏ3๐คฉ2๐1
Perplexity released an agentic Voice Assistant
It uses web browsing and multi-app actions to book reservations, send emails and calendar invites, play podcasts/videos, and more
Currently available in the Perplexity app, but only on iOS
It uses web browsing and multi-app actions to book reservations, send emails and calendar invites, play podcasts/videos, and more
Currently available in the Perplexity app, but only on iOS
โก๏ธViral rumors of DeepSeek R2 leaked
โ1.2T param, 78B active, hybrid MoE
โ97.3% cheaper than GPT 4o ($0.07/M in, $0.27/M out)
โ5.2PB training data. 89.7% on C-Eval2.0
โBetter vision. 92.4% on COCO
โ82% utilization in Huawei Ascend 910B
Big shift away from US supply chain.
โ1.2T param, 78B active, hybrid MoE
โ97.3% cheaper than GPT 4o ($0.07/M in, $0.27/M out)
โ5.2PB training data. 89.7% on C-Eval2.0
โBetter vision. 92.4% on COCO
โ82% utilization in Huawei Ascend 910B
Big shift away from US supply chain.
A new paper from Google DeepMind shows how Reinforcement Learning Fine-Tuning (RLFT) on self-generated Chain-of-Thought (CoT) can improve exploration and decision-making.
RLFT Implementation:
1. Set up the LLM to interact with a decision environment (e.g., bandit, Tic-Tac-Toe).
2. Prompt the LLM to generate a thinking process (CoT) and an action.
3. Extract and execute the action in the environment to get a reward.
4. Use the reward to fine-tune the LLM (via RL) based on the generated CoT and action.
5. Repeat interactions and fine-tuning to improve the LLM's decision policy.
Insights:
-LLMs act greedily, sticking to early successful actions and failing to explore potentially better options.
- Smaller LLMs tend to repeat actions common in the prompt history.
- LLMs can often articulate or calculate the correct strategy, but fail to execute
- Simple Reward bonuses (e.g., +1 for exploring a new action) or penalties (e.g., -5 for an invalid action format) guide LLM towards desired behaviour.
- Thinking Time Matters, allowing more tokens for generation improves performance
- Larger models suffer less frequency bias but are still prone to greediness
RLFT Implementation:
1. Set up the LLM to interact with a decision environment (e.g., bandit, Tic-Tac-Toe).
2. Prompt the LLM to generate a thinking process (CoT) and an action.
3. Extract and execute the action in the environment to get a reward.
4. Use the reward to fine-tune the LLM (via RL) based on the generated CoT and action.
5. Repeat interactions and fine-tuning to improve the LLM's decision policy.
Insights:
-LLMs act greedily, sticking to early successful actions and failing to explore potentially better options.
- Smaller LLMs tend to repeat actions common in the prompt history.
- LLMs can often articulate or calculate the correct strategy, but fail to execute
- Simple Reward bonuses (e.g., +1 for exploring a new action) or penalties (e.g., -5 for an invalid action format) guide LLM towards desired behaviour.
- Thinking Time Matters, allowing more tokens for generation improves performance
- Larger models suffer less frequency bias but are still prone to greediness
๐8โค4๐2
HuggingFace introduced SO-101 the new version of the hugely popular SO-100 low-cost robot arm:
- easier to assemble
- more robust in daily use
- still 100% open-source
- still ultra low-cost
Wowrobot shop
Seeedstudio shop
Partabot shop
- easier to assemble
- more robust in daily use
- still 100% open-source
- still ultra low-cost
Wowrobot shop
Seeedstudio shop
Partabot shop
๐ฅ7โค3๐2
Mastercard announced the launch of a global stablecoin payment system, covering wallet enablement, card issuance, merchant settlement, and on-chain remittances, and will partner with OKX to issue the OKX Card, linking crypto trading with everyday spending.
Mastercard is also collaborating with Circle, Nuvei, and Paxos to enable direct merchant settlement in stablecoins.
Mastercard is also collaborating with Circle, Nuvei, and Paxos to enable direct merchant settlement in stablecoins.
Coindesk
Mastercard Moves To Make Stablecoins Easier To Spend, Launches Crypto Card With OKX
Mastercard's new global system aims to make stablecoin transactions as seamless as traditional payments.
๐ฅ4โค3๐2
OpenAI is introducing shopping features in ChatGPT today, powered by GPT-4o, for ChatGPT Pro, Plus, and Free users, as well as logged-out users worldwide.
TechCrunch
OpenAI upgrades ChatGPT search with shopping features | TechCrunch
OpenAI is updating ChatGPT Search to give users an improved shopping experience, the company announced in a blog post.
๐ฅ4โค3๐2
Alibaba Introduced Qwen3
Open-weight Qwen3, latest large language models, including 2 MoE models and 6 dense models, ranging from 0.6B to 235B.
Qwen3-235B-A22B, achieves competitive results in benchmark evaluations of coding, math, general capabilities, etc., when compared to other top-tier models such as DeepSeek-R1, o1, o3-mini, Grok-3, and Gemini-2.5-Pro.
Additionally, the small MoE model, Qwen3-30B-A3B, outcompetes QwQ-32B with 10 times of activated parameters, and even a tiny model like Qwen3-4B can rival the performance of Qwen2.5-72B-Instruct.
Trained on 36T tokens, covering 119 languages! Data extracted from PDFs, synthetic data, etc.
Thinking and non-thinking modes
Improved agentic, coding capabilities, support for MCP
Training pipeline similar to DeepSeek R1
Small distilled models, such as Qwen3-4B that can rival the performance of Qwen2.5-72B-Instruct, even a Qwen3-0.6B model
GitHub
HuggingFace
Modelscope
Open-weight Qwen3, latest large language models, including 2 MoE models and 6 dense models, ranging from 0.6B to 235B.
Qwen3-235B-A22B, achieves competitive results in benchmark evaluations of coding, math, general capabilities, etc., when compared to other top-tier models such as DeepSeek-R1, o1, o3-mini, Grok-3, and Gemini-2.5-Pro.
Additionally, the small MoE model, Qwen3-30B-A3B, outcompetes QwQ-32B with 10 times of activated parameters, and even a tiny model like Qwen3-4B can rival the performance of Qwen2.5-72B-Instruct.
Trained on 36T tokens, covering 119 languages! Data extracted from PDFs, synthetic data, etc.
Thinking and non-thinking modes
Improved agentic, coding capabilities, support for MCP
Training pipeline similar to DeepSeek R1
Small distilled models, such as Qwen3-4B that can rival the performance of Qwen2.5-72B-Instruct, even a Qwen3-0.6B model
GitHub
HuggingFace
Modelscope
Qwen
Qwen3: Think Deeper, Act Faster
QWEN CHAT GitHub Hugging Face ModelScope Kaggle DEMO DISCORD
Introduction Today, we are excited to announce the release of Qwen3, the latest addition to the Qwen family of large language models. Our flagship model, Qwen3-235B-A22B, achieves competitive resultsโฆ
Introduction Today, we are excited to announce the release of Qwen3, the latest addition to the Qwen family of large language models. Our flagship model, Qwen3-235B-A22B, achieves competitive resultsโฆ
๐4โค3๐2
New work on automated prompt engineering for personalized text-to-image generation:
PRISM: Automated Black-box Prompt Engineering for Personalized Text-to-Image Generation
Paper + Code
Prompt engineering for personalized image generation is labor-intensive or requires model-specific tuning, limiting generalization.
Key Idea: PRISM uses VLMs and iterative in-context learning to automatically generate effective, human-readable prompts using only black-box access to image generation models.
This approach shows strong generalization and versatility in generating accurate prompts for objects, styles and images across multiple T2I models, including Stable Diffusion, DALL-E, and Midjourney. It also enables easy editing and multi-concept prompt generation.
PRISM: Automated Black-box Prompt Engineering for Personalized Text-to-Image Generation
Paper + Code
Prompt engineering for personalized image generation is labor-intensive or requires model-specific tuning, limiting generalization.
Key Idea: PRISM uses VLMs and iterative in-context learning to automatically generate effective, human-readable prompts using only black-box access to image generation models.
This approach shows strong generalization and versatility in generating accurate prompts for objects, styles and images across multiple T2I models, including Stable Diffusion, DALL-E, and Midjourney. It also enables easy editing and multi-concept prompt generation.
kellyyutonghe.github.io
PRISM: Automated Black-box Prompt Engineering for Personalized Text-to-Image Generation
We propose an algorithm that automatically identifies human-interpretable and transferable prompts that can effectively generate desired concepts given only black-box access to T2I models.
BCG_AI_Agents_MCP_1745919815.pdf
22.8 MB
BCG ๐ฑ๐ฟ๐ผ๐ฝ๐ฝ๐ฒ๐ฑ ๐๐ต๐ฒ๐ถ๐ฟ ๐น๐ฎ๐๐ฒ๐๐ ๐ฃ๐ข๐ฉ ๐ผ๐ป ๐๐ ๐๐ด๐ฒ๐ป๐๐ ๐ฎ๐ป๐ฑ ๐๐ต๐ฒ ๐ ๐ผ๐ฑ๐ฒ๐น ๐๐ผ๐ป๐๐ฒ๐
๐ ๐ฃ๐ฟ๐ผ๐๐ผ๐ฐ๐ผ๐น (๐ ๐๐ฃ)
๐๐ฒ๐ฟ๐ฒ ๐ฎ๐ฟ๐ฒ ๐ธ๐ฒ๐ ๐๐ฎ๐ธ๐ฒ๐ฎ๐๐ฎ๐๐:
1. ๐๐๐๐ผ๐ป๐ผ๐บ๐ผ๐๐ ๐๐ด๐ฒ๐ป๐๐ ๐๐ฟ๐ฒ ๐ ๐ผ๐๐ถ๐ป๐ด ๐๐ฟ๐ผ๐บ ๐๐ผ๐ป๐ฐ๐ฒ๐ฝ๐ ๐๐ผ ๐ฅ๐ฒ๐ฎ๐น๐ถ๐๐:
โ Early deployments are already delivering 30โ90% improvements in speed, productivity, and cost across coding, compliance, and supply chain domains.
2. ๐ ๐๐ฃ ๐๐ ๐๐ฒ๐ฐ๐ผ๐บ๐ถ๐ป๐ด ๐๐ต๐ฒ ๐๐ฎ๐ฐ๐ธ๐ฏ๐ผ๐ป๐ฒ ๐ผ๐ณ ๐ฆ๐ฐ๐ฎ๐น๐ฎ๐ฏ๐น๐ฒ ๐๐ด๐ฒ๐ป๐๐:
โ The Model Context Protocol (MCP) is the new open standard adopted by Anthropic, OpenAI, Microsoft, Google, and Amazon to expose tools, prompts, and resources reliably.
3. ๐๐ด๐ฒ๐ป๐ ๐๐ป๐๐ฒ๐น๐น๐ถ๐ด๐ฒ๐ป๐ฐ๐ฒ ๐๐ ๐ฃ๐ฟ๐ผ๐ด๐ฟ๐ฒ๐๐๐ถ๐ป๐ด ๐ฅ๐ฎ๐ฝ๐ถ๐ฑ๐น๐:
โ Agents today can automate tasks up to one hour long โ and this limit is doubling every seven months, pushing toward multi-day autonomous workflows by the end of the decade.
4. ๐๐ด๐ฒ๐ป๐ ๐๐ฟ๐ฐ๐ต๐ถ๐๐ฒ๐ฐ๐๐๐ฟ๐ฒ๐ ๐ ๐๐๐ ๐๐ฒ ๐ฆ๐ฒ๐ฐ๐๐ฟ๐ถ๐๐-๐๐ถ๐ฟ๐๐:
โ Security challenges grow as agents gain system access. OAuth, RBAC, permission isolation, eval-driven development, and real-time monitoring are mandatory to deploy agents safely.
5. ๐ง๐ต๐ฒ ๐ฅ๐ถ๐๐ฒ ๐ผ๐ณ ๐๐ด๐ฒ๐ป๐-๐ข๐ฟ๐ฐ๐ต๐ฒ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป ๐ฃ๐น๐ฎ๐๐ณ๐ผ๐ฟ๐บ๐:
โ Platforms like Azure Foundry, Vertex AI, Bedrock Agents, and Lindy are positioning themselves as the orchestration layer to create, manage, and scale enterprise agent ecosystems.
6. ๐๐ฟ๐ผ๐บ ๐ช๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐๐ ๐๐ผ ๐๐๐น๐น๐ ๐๐๐๐ผ๐ป๐ผ๐บ๐ผ๐๐ ๐๐ด๐ฒ๐ป๐๐:
โ Enterprises are shifting from prompt chaining (rigid workflows) to fully autonomous agents capable of observing, reasoning, and acting dynamically based on real-world feedback.
7. ๐ ๐๐ฃ ๐ฎ๐ป๐ฑ ๐2๐ ๐ช๐ถ๐น๐น ๐๐ฒ๐ณ๐ถ๐ป๐ฒ ๐๐ต๐ฒ ๐๐ด๐ฒ๐ป๐ ๐๐ฐ๐ผ๐ป๐ผ๐บ๐:
โ MCP connects agents to tools and data. A2A (Agent-to-Agent communication) will enable agents to negotiate, collaborate, and coordinate across systems โ forming true multi-agent networks.
๐๐ฒ๐ฟ๐ฒ ๐ฎ๐ฟ๐ฒ ๐ธ๐ฒ๐ ๐๐ฎ๐ธ๐ฒ๐ฎ๐๐ฎ๐๐:
1. ๐๐๐๐ผ๐ป๐ผ๐บ๐ผ๐๐ ๐๐ด๐ฒ๐ป๐๐ ๐๐ฟ๐ฒ ๐ ๐ผ๐๐ถ๐ป๐ด ๐๐ฟ๐ผ๐บ ๐๐ผ๐ป๐ฐ๐ฒ๐ฝ๐ ๐๐ผ ๐ฅ๐ฒ๐ฎ๐น๐ถ๐๐:
โ Early deployments are already delivering 30โ90% improvements in speed, productivity, and cost across coding, compliance, and supply chain domains.
2. ๐ ๐๐ฃ ๐๐ ๐๐ฒ๐ฐ๐ผ๐บ๐ถ๐ป๐ด ๐๐ต๐ฒ ๐๐ฎ๐ฐ๐ธ๐ฏ๐ผ๐ป๐ฒ ๐ผ๐ณ ๐ฆ๐ฐ๐ฎ๐น๐ฎ๐ฏ๐น๐ฒ ๐๐ด๐ฒ๐ป๐๐:
โ The Model Context Protocol (MCP) is the new open standard adopted by Anthropic, OpenAI, Microsoft, Google, and Amazon to expose tools, prompts, and resources reliably.
3. ๐๐ด๐ฒ๐ป๐ ๐๐ป๐๐ฒ๐น๐น๐ถ๐ด๐ฒ๐ป๐ฐ๐ฒ ๐๐ ๐ฃ๐ฟ๐ผ๐ด๐ฟ๐ฒ๐๐๐ถ๐ป๐ด ๐ฅ๐ฎ๐ฝ๐ถ๐ฑ๐น๐:
โ Agents today can automate tasks up to one hour long โ and this limit is doubling every seven months, pushing toward multi-day autonomous workflows by the end of the decade.
4. ๐๐ด๐ฒ๐ป๐ ๐๐ฟ๐ฐ๐ต๐ถ๐๐ฒ๐ฐ๐๐๐ฟ๐ฒ๐ ๐ ๐๐๐ ๐๐ฒ ๐ฆ๐ฒ๐ฐ๐๐ฟ๐ถ๐๐-๐๐ถ๐ฟ๐๐:
โ Security challenges grow as agents gain system access. OAuth, RBAC, permission isolation, eval-driven development, and real-time monitoring are mandatory to deploy agents safely.
5. ๐ง๐ต๐ฒ ๐ฅ๐ถ๐๐ฒ ๐ผ๐ณ ๐๐ด๐ฒ๐ป๐-๐ข๐ฟ๐ฐ๐ต๐ฒ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป ๐ฃ๐น๐ฎ๐๐ณ๐ผ๐ฟ๐บ๐:
โ Platforms like Azure Foundry, Vertex AI, Bedrock Agents, and Lindy are positioning themselves as the orchestration layer to create, manage, and scale enterprise agent ecosystems.
6. ๐๐ฟ๐ผ๐บ ๐ช๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐๐ ๐๐ผ ๐๐๐น๐น๐ ๐๐๐๐ผ๐ป๐ผ๐บ๐ผ๐๐ ๐๐ด๐ฒ๐ป๐๐:
โ Enterprises are shifting from prompt chaining (rigid workflows) to fully autonomous agents capable of observing, reasoning, and acting dynamically based on real-world feedback.
7. ๐ ๐๐ฃ ๐ฎ๐ป๐ฑ ๐2๐ ๐ช๐ถ๐น๐น ๐๐ฒ๐ณ๐ถ๐ป๐ฒ ๐๐ต๐ฒ ๐๐ด๐ฒ๐ป๐ ๐๐ฐ๐ผ๐ป๐ผ๐บ๐:
โ MCP connects agents to tools and data. A2A (Agent-to-Agent communication) will enable agents to negotiate, collaborate, and coordinate across systems โ forming true multi-agent networks.
๐3๐ฅ2