The day in one read

The AI infrastructure trade faced a sharp reality check on October 15, as a hotter-than-expected inflation print pushed the probability of a December Federal Reserve rate hike to 87% and triggered a broad sell-off in technology stocks. The S&P 500 fell 1.1% to 7,719 points, ending a six-year streak without a daily loss exceeding 1%, while the Nasdaq 100 dropped 2.0% to 29,950 points. The correction was particularly severe for AI-centric names, with the Neo-Cloud group losing 12% and semiconductor stocks marking their sixth consecutive loss day. This market turbulence highlights a growing divergence between the explosive revenue growth of AI model providers and the strained credit markets underpinning the physical infrastructure required to support them.

The AI Credit Crunch and Market Correction

The bond market is emerging as a more honest early indicator of AI sector risk than equity prices, with credit spreads widening significantly for neocloud providers. CoreWeave’s stock dropped over 10% after its bond yield reached 15%, a level that signals distressed credit conditions despite the company’s equity gains. This sentiment dragged down peers Iron and Nebius, which fell 12.2% and 10% respectively. The pressure extended to the semiconductor supply chain, where Nvidia fell 3.2%, AMD dropped 5.7%, and Marvell lost 8.7%. ASML shares fell nearly 4% in New York despite reporting extreme demand from customers like TSMC and Intel, with orders extending to 2029 and 2030. Analysts note that ASML’s production constraints are capping its valuation, as the company explores production increases beyond its standard 30% annual target.

The macro backdrop intensified the sell-off, with US consumer prices rising 3.7% year-over-year in September, exceeding the 3.6% forecast. The 10-year US Treasury yield ended at 5.19%, reflecting the market’s pricing of tighter monetary policy. In response to the volatility, investors rotated toward defensive assets and established giants. Apple became the second company to reach a $5 trillion market cap, while Berkshire Hathaway increased its stake in Bending Spoons to 4% by investing over $300 million at a discount, a move that sent the Italian tech firm’s stock surging 50%. Meanwhile, Bitcoin fell 1.9% to $81,800, and gold dropped 1.4% to $4,100, suggesting a broad risk-off environment rather than a flight to specific safe havens.

Agentic Infrastructure and the Trust Layer

As AI agents move from experimental tools to production systems, the industry is grappling with the need for robust trust layers and standardized infrastructure. G2’s Chief Product Officer Alexis Zhang argues that effective agentic AI requires a framework combining verification via evaluations, guardrails to block incorrect actions, and observability to track deployed systems. She emphasizes that product managers must shift from deterministic roadmaps to probabilistic convergence strategies, noting that AI-generated products often converge on common traits due to training data biases. This challenge is mirrored in the development of webhook infrastructure, where Sphix argues that the lack of standardization is a critical bottleneck. Sphix co-created the "standard webhooks" spec, adopted by OpenAI, Anthropic, and Google Gemini, to address the "Jevon's paradox" where ease of use drives increased complexity. The company is also developing "polling endpoints" to handle the ephemeral nature of local agent environments, preventing the "storm of retries" that occurs when local connections drop.

Enterprise adoption of these agentic systems is accelerating, but it requires centralized governance to manage cost and security. Delivery Hero’s SVP of Tech Foundations, Rodrigo, detailed the company’s transition to an autonomous software engineering model branded as HeroGen. In Q1, the team targeted 100 merged pull requests but achieved 1,800, an 18x overage, with an 83% merge rate by September. To manage this scale, Delivery Hero centralized its agentic tooling stack for its 1,500 Berlin-based engineers, moving away from decentralized experimentation. The company uses Claude as its primary tool, with subsidiaries in regions like South America using Gemini. Rodrigo predicts that by the end of next year, the system will handle end-to-end production deployments safely, including automated incident handling. Similarly, Plio, a SaaS CRM specialized in retail, is transitioning its engineering strategy toward an agentic AI platform. The company launched an "Ask CRM" agent that reduced customer analysis time from three weeks to a few prompts, and is now preparing for external integration via MCP to allow customers to connect Plio’s data to their own external agents.

The SaaS Consolidation and Venture Landscape

The SaaS market remains under significant pressure, with mid-market firms facing a downturn described as a "worse Saspocalypse" than in early 2026. While large-cap SaaS companies have recovered, the high bar for IPOs and acquisition targets has left many mid-sized firms struggling. Bending Spoons is laying off most of Airtable’s staff post-acquisition, and Flock Safety is cutting 18% of its staff amid security and privacy backlash. Bumble is being taken private by Ithaca Holdings for $1.3 billion, a move that reflects the broader trend of consolidation in the sector. In the venture capital space, valuations are reaching historic highs, with Jev, the parent company of TypeSafe, raising an $870 million Series A at a $7.5 billion valuation, led by Andreessen Horowitz. This round is described as one of the largest Series A rounds ever, with usage quadrupling since the deal was struck.

Disputes over investor rights are also emerging as valuations soar. Valor Atomics is seeking a declaratory judgment against seed investor Day One Ventures regarding pro-rata rights in its $1 billion Series B, valued at $6 billion and led by Sequoia. The dispute centers on Sequoia’s requirement that existing investors waive pro-rata rights to meet ownership targets, a practice Day One contests via springback clauses. Meanwhile, Meta released Muse Wonder, a closed-source model performing just shy of Opus 5.5 on independent indices, marking its return to the frontier in closed-source AI. Rumors also indicate that NVIDIA is acquiring open-source model developer Reflection for $30 billion, following its $12 billion acquisition of Poolside, a move that aligns with its strategy to capture value in the open-weight model space.

Agentic Commerce and Payment Infrastructure

The rise of AI agents is reshaping the payment landscape, creating a new "wallet wars" dynamic where platforms compete to connect with agents. VGS, which secures payment credentials for approximately one-third of all e-commerce GMV, is transitioning from a static vault to a "network era" role, brokering credentials between consumers, AI agents, and merchants. The company is the largest provisioner of "agentic tokens," issuing millions weekly to allow agents to transact without exposing raw card numbers. This infrastructure is critical as wallets like Stripe Link, with 300 million consumers, and PayPal, with 400 million accounts, rush to integrate with agents. However, the shift is not uniform; Amazon recently blocked the Muse agent, likely to protect its advertising revenue model, while Walmart and small businesses may welcome agents as new channels.

The infrastructure must also handle the complexities of multi-processor environments, where merchants use 3-5 payment processors to optimize costs. VGS notes that about 25% of cards turn over annually due to expiration or network changes, such as Capital One’s acquisition of Discover. As agents evolve from spectators to actors, capable of executing complex tasks like booking services, new liability questions regarding "delegated authority" and agent errors are emerging. The card networks have introduced open standards for agentic tokens to solve the "walled garden" approach used by some platforms, which complicates chargebacks and follow-on transactions like subscriptions.

Content Moderation and Corporate Restructuring

The proliferation of AI-generated content is forcing media and entertainment companies to rethink their moderation strategies and resource allocation. Deezer announced a new toggle feature, expected before year-end, allowing users to filter out AI-generated music, albums, and artists from search results. CEO Alexis Lanthornier cited a survey showing 45% of customers wanted this option, noting that AI-generated tracks rose from 10% of daily uploads last year to over 50% in July, averaging 90,000 tracks per day. Competitors have adopted varying policies, with Bandcamp banning AI tracks and Spotify tagging AI content prominently. Meta launched "Your Algorithm" on Facebook, an AI-powered feature enabling users to influence feed recommendations via natural language descriptions, aiming to make recommendations more relevant as AI shapes content discovery.

Netflix confirmed layoffs affecting approximately 1,000 employees, or 6% of its 16,000 full-time staff, primarily in product, tech, legal, and HR departments. Co-CEOs Ted Sarandos and Greg Peters stated in an internal memo that the company must be disciplined in resource allocation, a move that follows Netflix's loss in the bidding war for Warner Bros. Discovery assets to Paramount's Skydance. The company faces intensifying competition from social platforms, with reports indicating YouTube surpassed Netflix in average daily viewing in 2025. Despite the cuts, Netflix continues investing in live events, sports, gaming, and AI-supported features to retain subscribers, having reported $45.2 billion in revenue in 2025, with ad revenue exceeding $1.5 billion.

Also Notable

Lufthansa won the bid for 45% of TAP Air Portugal for over 1 billion euros, potentially up to 1.2 billion, beating Air France-KLM in a move that marks Lufthansa’s return to the DAX after six years. Bank of America shares fell 5% following a fine of over $1 billion for money laundering violations and weak earnings, with equity trading growth at just 10% compared to 70% at Goldman Sachs. BlackRock attracted $205 billion in net new capital, managing $15.5 trillion in assets, while Morgan Stanley gained $232 billion in wealth management assets. Corgi launched the MANGO ETF holding $75 million in Meta, Nvidia, Google, and SpaceX, using swaps to access perpetual futures for private firms Anthropic and OpenAI, a structure that has drawn criticism for hindering direct capital raising for those firms. Mozilla CTO Rafi outlined a strategy to democratize AI by preventing the concentration of intelligence in a few large tech companies, advocating for a decentralized ecosystem where users own their data and tools rather than renting them from centralized providers.