SOON is gearing up to launch a perpetual trading platform built specifically for financial AI agents, with a target go-live date of September 30, 2026. The platform, called 100xSOON, already went live on BNB Chain, offering up to 10,000x leverage across crypto, stocks, and metals. Tradable assets include BTC, ETH, SOL, and stocks like NVDA, TSLA, and AAPL, along with gold and silver. Trading is exclusive to the $SOON token, and 80% of platform revenue funds buybacks and burns of that token. Positions settle automatically every 30 seconds, and agents connect via the MCP protocol while paying through the x402 payment protocol.
Key Takeaways
- <ul><li>100xSOON launched on BNB Chain, offering up to 10,000x leverage for AI agents trading crypto, stocks, and metals</li><li>$SOON token surged 31% after SOON invested in Phala Network's compute infrastructure for AI agents</li><li>HPE raised networking growth outlook to high-teens CAGR and landed a $1.2 billion AMD server order from Vultr</li><li>ElevenLabs doubled its valuation to $22 billion through a $300 million tender offer</li><li>AdviSD trains small advisors to steer frontier LLMs like Gemini and Claude more effectively</li><li>NVIDIA released a NeMo Relay tutorial for tracing agent harness behavior during evaluation</li><li>Genea launched Genea MCP, an AI tool for managing building access through natural language commands</li><li>Park Place introduced ParkView Asset Intelligence, scoring IT hardware health from 0 to 100</li><li>Trump announced tech CEOs signed a voluntary AI self-regulation accord with no legal force</li><li>Economists note AI's broad economic impact remains slow despite rising solopreneur activity</li></ul>
SOON launches AI agent perpetual trading platform by September 2026
SOON is preparing a perpetual trading platform built for financial AI agents, scheduled to go live by September 30, 2026. The platform will provide agent-native MCP access so AI agents can connect and trade directly. Trading will be exclusive to the $SOON token, and revenue generated will be used for buyback and burn of $SOON. The effectiveness depends on undisclosed factors like the burn percentage and actual post-launch usage by agents.
100xSOON launches high-leverage AI trading on BNB Chain
100xSOON went live as a high-leverage trading platform on BNB Chain, offering up to 10,000x leverage in crypto, stocks, and metals with $SOON token settlement. The platform is built for financial AI agents, allowing them to connect through the MCP protocol and pay via the x402 payment protocol using $SOON. Tradable assets include BTC, ETH, SOL, stocks like NVDA, TSLA, AAPL, and CRCL, plus gold and silver. 80% of platform revenue goes toward buying back and burning $SOON tokens, and positions settle automatically every 30 seconds.
SOON token jumps 31% on AI agent plans and Phala investment
SOON surged 31% after investing in Phala Network's compute infrastructure, which provides the computing layer for AI agents in its ecosystem. The company also teased a perpetual trading platform for financial AI agents, where trading will be exclusive to $SOON and revenue will fund buyback and burn of the token. A September 23 unlock released about 20.24 million SOON tokens, with another similar unlock scheduled for October 23. SOON moved from roughly $0.19 earlier in September to around $0.45, driven by AI-agent infrastructure and derivatives activity.
HPE Investor Day expected to highlight AMD AI opportunity
Hewlett Packard Enterprise (HPE) is hosting an investor day on Wednesday, with shares surging 160% in 2026 due to upbeat views on AI infrastructure build-outs. The event is expected to focus on HPE's networking business following the $14 billion acquisition of Juniper Networks in 2025. Rami Rahim, executive VP and general manager of HPE's Intelligent Edge, will lead the discussion on the company's AI strategy.
HPE raises networking growth outlook, lands $1.2 billion AI order from Vultr
Hewlett Packard Enterprise raised its long-term revenue growth forecast for its networking business and announced a $1.2 billion order to provide AMD server racks to cloud firm Vultr. HPE now expects networking segment revenue to grow at a compound annual growth rate in the high teens from fiscal 2026 to fiscal 2029, up from its previous forecast of 5% to 7%. The company will provide AMD Helios AI racks, equipped with AI chips and HPE's networking switches and software, to Vultr's US data centers.
Trump backs voluntary AI self-regulation as tech CEOs sign accord
President Donald Trump announced that tech executives signed a voluntary agreement to self-regulate artificial intelligence. Trump praised the accord as a positive step toward self-policing. The agreement has no legal force and signatories are not required to follow it. Critics argue self-regulation is not enough to prevent dangerous AI mistakes. Some experts warned that AI models can lie or cheat even when they know better.
New risk-aware planning method improves POMDP safety guarantees
Researchers proposed a new method that applies Conditional Value at Risk to immediate costs in online POMDP planning. The approach targets uncertainty at each step instead of modifying the value function. It keeps the standard expected cumulative return as the objective. This allows existing expectation-based planners to become risk-sensitive with only small changes. The method provides finite-time guarantees and recovers standard planning in the risk-neutral limit.
AdviSD helps small advisors steer frontier LLMs more effectively
AdviSD is a new method that trains small advisors to guide large language models using natural-language advice. It pairs reinforcement learning with selective self-distillation from feedback. The method avoids learning from every correction, which can limit improvement. Experiments with Qwen3-8B advisors for Gemini and Claude showed better performance on BFCL-v3 and EnvScaler. The trained advisors also generalized to new tasks and model families.
NVIDIA NeMo Relay traces agent harness behavior for better evaluation
A NVIDIA developer tutorial explains how to use execution traces with NeMo Relay to study agent behavior. It shows how agents can finish tasks while taking inefficient paths or repeating steps. The tutorial uses two Hermes Agent examples, including a terminal task and a multi-tool research task. It recommends comparing baseline and candidate harness changes with fixed tasks and automated success checks. Traces help examine model calls, tool usage, errors, timing, tokens, and cost.
ElevenLabs doubles valuation to $22 billion in new tender offer
AI voice startup ElevenLabs reached a $22 billion valuation after a $300 million tender offer. The offer allowed employees to sell shares to investors led by Wellington and T. Rowe Price. ElevenLabs was founded in 2022 and is known for realistic voice and sound effect generation. The company is based in New York and London and is now among Europe's most valuable startups.
AI Promises Big Gains but Real-World Impact Remains Slow
AI has the potential to boost worker productivity, raise wages, and lower prices, but the financial benefits arrive slowly. Economists find that despite rapid advances in AI capabilities, the broad economic impact so far has been minimal. One positive sign is a rise in new business formation, especially among solo entrepreneurs called solopreneurs. Patience is still needed as companies and institutions work out how to use these tools effectively.
Genea Launches AI Tool for Managing Building Access
Genea released Genea MCP, a Model Context Protocol server that lets AI agents handle access control tasks. The tool lets administrators grant or change access by speaking naturally, such as telling it to give someone entry to a specific floor. It inherits existing user permissions and requires administrator approval to turn on. Before any high-impact action like deleting a user, it shows what will be affected and waits for confirmation.
Park Place Launches AI Platform for Hardware Health Insights
Park Place Technologies introduced ParkView Asset Intelligence, an AI-powered platform that checks the health of IT hardware. It analyzes maintenance and service data to give each device a lifecycle health score from 0 to 100. The tool helps IT teams decide whether to refresh, repair, or retire hardware, including older systems and new AI infrastructure. Customers can use the scores to justify keeping equipment running or to plan maintenance budgets.
Sources
- SOON: Agent Perpetual Trading Platform for Financial AI Agents Goes Live
- High Leverage AI Trading Launches on 100xSOON Platform
- SOON Jumps 31% as AI Agent Plans Fuel a Fresh Rally
- Will HPE Investor Day Put AMD AI Opportunity In Spotlight?
- HPE boosts networking growth outlook, gets $1.2 billion AI order from cloud firm Vultr
- Trump Rejects AI Regulation as Tech CEOs Sign Voluntary “Self-Policing” Accord
- Risk-Averse Online POMDP Planning via CVaR of the Immediate Cost with Performance Guarantees
- AdviSD: Learning to Advise Frontier LLMs via Targeted Multi-Turn Self-Distillation
- Tracing Agent Harness Behavior with NVIDIA NeMo Relay
- AI voice startup ElevenLabs doubles valuation to $22B
- Artificial Intelligence Meets Real-World Frictions
- Genea brings AI agents to access control with role-based permissions
- Park Place’s New AI Platform Gives OEM-Level Hardware Visibility And Insights: Exclusive
Comments
Please log in to post a comment.