Morgan Stanley’s E*Trade is in discussions to take a leading role in distributing shares to retail investors in SpaceX’s upcoming initial public offering, according to the report from Reuters citing people familiar with the matter.

The move could give E*Trade a significant advantage over rival platforms such as Robinhood and SoFi, which have also sought participation in the deal. SpaceX is reportedly considering limiting or excluding those firms from the retail allocation, an unusual step given their growing presence in major IPOs in recent years.

The SpaceX listing is expected to be the largest in history, with strong demand anticipated from both institutional and individual investors.

Retail Access Becomes Strategic Battleground

Retail investors are expected to play a larger role than usual in the SpaceX IPO. The company is reportedly considering allocating up to 30% of shares to individual investors, well above the typical 5% to 10% seen in most public offerings.

Morgan Stanley, a lead underwriter on the deal, is expected to channel a significant portion of that allocation through its E*Trade platform. This strategy would allow the bank to capture more of the retail order flow internally, rather than relying on third-party brokerages.

Robinhood and SoFi remain in discussions but may receive a smaller share of the offering, if any. Fidelity is also reportedly seeking a role in distributing shares through its platform.

The plans are still under discussion and could change as the IPO approaches.

Morgan Stanley’s Push Into Retail

A prominent role in the SpaceX IPO would mark a major win for E*Trade, which Morgan Stanley acquired for $13 billion in 2020. The bank has since expanded its focus on retail trading as part of a broader strategy to diversify revenue beyond traditional investment banking and wealth management.

Securing a central position in a high-profile IPO could strengthen E*Trade’s competitive standing against platforms such as Charles Schwab and Interactive Brokers, particularly as retail participation in equity markets continues to grow.

The allocation strategy also reflects a broader shift in how IPOs are structured, with increased emphasis on engaging individual investors alongside institutional buyers.

Implications for the IPO Market

The SpaceX IPO is shaping up to be a landmark event, not only because of its scale but also due to its unconventional structure and strong retail focus. The involvement of platforms like E*Trade highlights how distribution strategies are evolving in response to changing investor dynamics.

The outcome could influence how future large-scale listings are structured, particularly for technology and AI-driven companies seeking to tap both institutional capital and retail enthusiasm.

As SpaceX moves closer to going public, decisions around share allocation and distribution will play a critical role in shaping demand and setting precedents for the next wave of high-profile IPOs.

AI & Machine Learning, News, Startups & Investment

OpenAI has introduced plugin support for Codex, expanding its development tool into a broader platform for AI-driven workflows with integrations across popular workplace applications.

The new feature allows users to connect Codex with services including Slack, Notion, Figma, Gmail, and Google Drive. Through these integrations, Codex can access external data, automate tasks, and execute workflows that extend beyond traditional code generation.

The launch also marks the beginning of a plugin marketplace strategy, where reusable AI workflows can be distributed and adopted across teams with minimal setup.

Building an AI Workflow Ecosystem

Plugins in Codex are designed as bundled units that combine predefined workflows, integrations with external applications, and support for Model Context Protocol servers. This structure allows developers and teams to create reusable configurations tailored to specific tasks.

For example, Codex can be used to summarize Slack channels, manage documents in Google Drive, or generate and modify designs through Figma integrations. These capabilities position the tool as more than a coding assistant, enabling it to function as a general-purpose productivity layer across enterprise environments.

Previously, similar workflows required manual configuration and technical expertise. With the introduction of plugins, users can install and deploy these capabilities through a centralized directory, lowering the barrier to adoption.

The approach aligns with a broader shift in the AI sector toward agent-based systems that can execute multi-step tasks across different tools and services.

Competing in the AI Platform Race

The expansion of Codex into a plugin-enabled platform reflects increasing competition among AI providers to build extensible ecosystems. Rivals have already emphasized integrations and modular architectures, particularly for enterprise use cases.

By launching a plugin marketplace, OpenAI is aiming to create a network effect around Codex, where third-party developers can contribute tools and workflows that enhance the platform’s capabilities. This model mirrors strategies seen in cloud software and developer platforms, where ecosystems play a key role in driving adoption.

The inclusion of widely used services such as Slack, Notion, and Gmail highlights a focus on real-world productivity use cases. It also signals a move toward embedding AI more deeply into everyday workflows, rather than limiting it to isolated development tasks.

As organizations increasingly adopt AI agents to automate complex processes, tools like Codex are evolving to serve as coordination layers across software environments. The addition of plugins positions OpenAI to capture a larger share of this emerging market for AI-powered work platforms.

AI & Machine Learning, News

DDR5 memory prices are beginning to show signs of easing after a prolonged surge driven by artificial intelligence demand, with analysts pointing to a recent breakthrough in AI efficiency as a key turning point.

The shift follows the introduction of TurboQuant, a compression algorithm unveiled by Google that significantly reduces the memory requirements of large AI models. By lowering demand for high-bandwidth memory and DRAM, the development is starting to rebalance a market that had been under intense pressure from AI infrastructure expansion.

The price movement marks a rare reversal after a sharp increase in 2025, when expectations around AI-driven demand pushed memory costs to record levels.

AI Demand Fueled Price Surge

Last year, the market reacted strongly to reports that OpenAI had signed preliminary agreements with major memory manufacturers Samsung and SK Hynix for up to 40% of global DRAM output. Although the agreements were non-binding letters of intent rather than firm purchase commitments, they were widely interpreted as indicative of massive future demand.

That perception drove DDR5 prices up by as much as 171%, with high-capacity memory kits becoming significantly more expensive. The surge also reflected broader investment in AI data centers, where memory is a critical component for training and running large-scale models.

However, some large infrastructure projects later faced delays or cancellations amid uncertainty over actual demand, contributing to growing volatility in the memory market.

TurboQuant Shifts Market Dynamics

The release of Google’s TurboQuant algorithm has introduced a new variable into the equation. The technology reduces key-value cache memory requirements by up to six times while maintaining performance, potentially lowering the amount of DRAM needed for AI workloads.

This improvement could have a direct impact on data center design, enabling operators to run large models with fewer memory modules. As a result, some supply may shift back toward consumer markets, including gaming and personal computing.

Early signs of this shift are emerging. In the United States, certain DDR5 modules, including Corsair Vengeance kits, have seen modest price declines at major retailers. Similar trends have been reported in parts of Europe, suggesting a broader stabilization in pricing.

Limited Relief for Consumers

Despite these developments, the overall memory market remains constrained. Most DRAM supply continues to be prioritized for enterprise customers, particularly hyperscalers building AI infrastructure.

Industry trackers indicate that while prices are leveling off, widespread declines have yet to materialize across all products. Analysts caution that improvements in efficiency could paradoxically drive further AI adoption, sustaining long-term demand for memory.

The broader impact of TurboQuant may depend on how quickly it is adopted and whether it leads to a net reduction in hardware requirements or enables even larger and more complex AI systems.

For now, the easing of DDR5 prices reflects an early adjustment in a market that has been heavily influenced by AI-driven expectations. It also highlights how advances in software efficiency can have immediate ripple effects across hardware supply chains.

AI & Machine Learning, Cloud & Infrastructure, Enterprise Tech, News