AI CEOs Slow Development; Ciena Targets $50B Market

Key Takeaways

• <ul><li>Paytm, PhonePe and Razorpay are launching AI products but expect no meaningful AI revenue until FY28 or later</li><li>Ciena targets a $50 billion addressable market by 2029 driven by AI networking demand</li><li>Chinese tech giants Baidu, Alibaba, Tencent and ByteDance are cutting AI token quotas to control costs</li><li>ByteDance caps annual AI token payouts at US$1,000 for technical staff; Alibaba caps external tool reimbursement at US$200 per month</li><li>Intel, Nvidia, Micron and SK Hynix stocks rebounded after Anthropic and OpenAI CEOs announced slower AI development for safety</li><li>SK Hynix broke ground on an HBM production base in Indiana on August 27, 2026, due to power grid constraints in South Korea</li><li>Apple Intelligence camera features require paid iCloud+ plans ranging from $10 to $60 per month depending on camera count</li><li>OpenAI CEO Sam Altman supports a federal framework for frontier AI safety requirements</li><li>Caliptra introduced hardware security IP designed specifically for AI SoC protection</li><li>A new AI framework using Large Language Models reduced EV charging optimization training time by over 70 percent</li></ul>

Paytm PhonePe Razorpay Build AI Products but Revenue Still Far Off

Paytm, PhonePe and Razorpay are launching AI-based products across fintech, but meaningful revenue from these tools is not expected until FY28 or later. Emkay Research said AI was the main theme at Global Fintech Fest 2026, with 32 percent of sessions focused on AI. None of the companies disclosed revenue from their new AI products. The near-term benefits are likely to come from lower costs and better customer service rather than direct income.

Fintech AI Revenue Likely Only From FY28 Says Emkay Research

India's fintech firms are rapidly building AI tools for payments, lending and customer service, but revenue from these products will likely start only from FY28. Emkay Research reported that 144 of 452 sessions at Global Fintech Fest 2026 focused on AI. Companies like Paytm, PhonePe and Razorpay demonstrated new AI tools but did not share any AI-specific revenue figures. The immediate gains are mainly operational, including lower service costs and faster technology integration.

Ciena Venture Fund Aims to Turn AI Networking Demand Into Growth

Ciena Corporation launched a venture program targeting four areas including networking infrastructure, next-generation data centers and optical networking. The company expects its addressable market to reach $50 billion by 2029 as AI drives demand for higher bandwidth. Fiscal third-quarter backlog rose to $8.5 billion and revenue guidance was raised to $6.42 billion. Fiscal 2027 revenue is expected to grow at least 30 percent.

Ciena Expands Venture Investments in AI Networking and Data Centers

Ciena's venture fund will invest in optical networking and interconnects, which match its core business as AI increases demand for connectivity. The company is benefiting from AI-driven network traffic across cloud and service providers. Fiscal 2027 revenue is expected to reach $8.3 to $8.4 billion with an adjusted operating margin of 25 to 27 percent. Ciena holds a Zacks Rank of 1, which is a Strong Buy rating.

New AI Model Optimizes EV Charging and Smart Grid Operations

Researchers proposed a new framework using Large Language Models and Multi-Agent Reinforcement Learning to optimize electric vehicle charging, charging stations and power grids together. The system uses real-time IoT data to select important features and balance competing goals like profit and user satisfaction. Experiments showed the framework reduced training time by over 70 percent compared to existing methods. The approach offers a scalable solution for managing urban EV charging infrastructure.

Caliptra security IP targets AI SoC protection

Caliptra introduced a hardware security solution designed for AI SoCs. The root-of-trust IP includes a complete hardware-software integration framework. It aims to orchestrate security for AI system-on-chip designs.

China tech giants limit AI token use at work

Employees at Baidu, Alibaba, Tencent and ByteDance face tighter AI token quotas. Companies are cutting back on computing allowances to control rising AI costs. ByteDance staff using closed-source models may co-pay, with annual payouts capped around US$1,000 for technical staff. Alibaba gives 3,000 to 6,000 monthly credits on its QoderWork platform and caps external tool reimbursement at US$200 per month. Baidu offers a 1,500 yuan monthly baseline, and Tencent now uses departmental token pools overseen by managers.

Semiconductor stocks rise after AI slowdown fears

Shares of Micron, Intel, SanDisk, Nvidia and SK Hynix climbed in early trading after a prior market drop. Intel rose 2.12% to $99.25 and SK Hynix gained 1.92%. The rebound followed announcements from Anthropic CEO Darius Amodei and OpenAI CEO Sam Altman about slowing AI development for safety. Altman also supported a federal framework for frontier AI safety requirements.

Power grid limits may slow AI memory expansion

South Korea's electricity grid cannot keep up with SK Hynix's planned AI memory factories. The country expects new semiconductor plants and data centers to add power demand equal to twenty nuclear reactors. KEPCO rejected a proposal requiring SK Hynix to prepay roughly five years of electricity costs. SK Hynix broke ground on an HBM production base in Indiana on August 27, 2026, to reduce grid and political risk.

Apple Home AI camera features carry high costs

Apple introduced Apple Intelligence security camera features for HomeKit Secure Video devices like Aqara cameras. The features include natural language search and descriptive video summaries but require paid iCloud+ plans. A single camera needs a $10-per-month 2TB plan, two cameras cost $30 per month, and up to five cameras cost $60 per month. The features are opt-in and available through iOS 27 settings.

AI security fears push cybersecurity stocks to new highs

Investors are buying cybersecurity stocks because they worry AI will power more cyber attacks. Experts point to deepfakes as one example of how AI makes attacks more dangerous. The cybersecurity sector has already grown a lot in recent years. Many stocks in the sector have more than doubled in value. The latest jump is being driven by AI-related concerns.

New method improves multi-objective optimization under tight budgets

Multi-objective Bayesian optimisation tries to find good solutions across many goals at once. The new converge-then-diversify approach splits the work into two stages. The first stage focuses on finding a strong point on the Pareto front. The second stage spreads solutions across the front for better diversity. Tests show it outperforms existing methods in about 73 percent of cases, especially when budgets are small.

Sources

• Paytm, PhonePe, Razorpay Are Building AI Products. When Will They Start Making Money?
• Fintech AI products to yield meaningful revenue only from FY28: Emkay
• CIEN's New Venture Fund: Can it Turn AI Networking Into Growth Engine?
• CIEN's New Venture Fund: Can it Turn AI Networking Into Growth Engine?
• LLM-Enhanced Multi-Agent Reinforcement Learning for Unified Electric Vehicles-Charging Station-Grid Optimization in Public Charging Systems
• Caliptra-enabled hardware security solution eyes AI SoCs
• Forget AI KPIs: how China’s tech giants are rationing tokens for employees
• AI 'Slowdown' Jitters Fading? Micron, Intel, SanDisk, Nvidia, SK Hynix Up Pre-Market Day After Crash
• What Stops the AI Memory Boom? It Might Have Nothing to Do With Chips.
• Apple Home’s AI Security Camera Features Are Opt-In, if You Can Afford Them
• AI fears send cybersecurity stocks higher
• Converge Then Diversify: Decoupling Convergence and Diversity in Multi-Objective Bayesian Optimisation