Tech Express | July 7, 2026 Samsung's $58.4 Billion Quarter Shatters All Tech Profit Records

---

📰 Samsung Posts Record $58.4 Billion Q2 Operating Profit on AI Memory Surge

Samsung Electronics released its Q2 2026 earnings guidance, projecting an operating profit of approximately 89.4 trillion won (about $58.4 billion), a staggering 1,810% year-over-year increase—the highest single-quarter profit ever recorded by any technology company. The explosive result was driven by insatiable demand for high-bandwidth memory (HBM) and DDR5 from AI server customers, cementing Samsung's position as the world's leading memory chipmaker.

The company has faced supply shortages for three consecutive quarters, with HBM production lines operating at full capacity. Meanwhile, AI chip giant NVIDIA continues to place massive orders for Samsung's HBM3E products, keeping Samsung deeply embedded in the AI infrastructure supply chain. Analysts expect AI memory demand to remain elevated through the second half of 2026.

The guidance significantly exceeded market expectations of 84.2 trillion won. Notably, Samsung's single-quarter profit this period surpassed any recent quarter from both NVIDIA and Apple, signaling the memory chip industry's full-scale resurgence in the AI era.

— The Eastern Herald

---

📰 Xiaomi Joins DeepSeek's AI Price War, Slashing API Prices by 99%

The Chinese AI model pricing war has escalated dramatically. Xiaomi announced sweeping cuts to its large model API pricing, slashing rates by up to 99% in a direct challenge to DeepSeek's earlier permanent price reductions. Under the new pricing, the entry-level plan includes 500 million tokens, directly disrupting the current domestic AI API market.

Over the past six months, China's AI model market has seen intense turbulence: DeepSeek opted for permanent price cuts, Alibaba suspended its low-cost Coding Plan Lite tier, ByteDance removed its budget Coding Plan, and Zhipu raised API pricing by 83% in Q1 2026. Xiaomi's aggressive entry aims to rapidly capture developer ecosystem share through ultra-low pricing.

Industry analysts note that while the API price war lowers barriers to AI application development in the short term, it may accelerate market consolidation. For startups, maintaining differentiation amid the pricing battle remains a critical challenge.

— 36Kr

---

📰 Tencent Officially Launches Hunyuan Hy3 AI Model with 20x Token Growth

Tencent officially released its next-generation AI model, Hunyuan Hy3, on July 6. The company states that Hy3 demonstrates significantly superior intelligence compared to models of similar size, with particularly notable improvements in productivity tasks including software development, office productivity, financial modeling, UI design, and game development. It rivals the performance of much larger flagship models.

Before its official launch, Hy3 was battle-tested through a preview version by global developers and across Tencent's vast real-world businesses, with daily token consumption growing 20-fold. Hy3 is already integrated into core Tencent products including WorkBuddy/CodeBuddy, Yuanbao, Marvis, and ima, with its API live on Tencent Cloud's TokenHub and additional overseas API platforms coming soon.

Remarkably, the Hunyuan team completed the entire R&D pipeline—from infrastructure rebuild to product feedback—in under six months, showcasing Tencent's rapid iteration capabilities in the AI foundation model space.

— Tencent

---

📰 Huawei Mate 90 to Feature New Kirin Chip Based on "Tao's Law"

According to informed sources cited by STAR Market Daily, Huawei's upcoming Mate 90 series, set to launch this fall, will debut a new Kirin chip built on "Tao's Law" principles. This marks the first smartphone chip to fully implement Tao's Law, promising a generational leap in performance.

In May 2026, Huawei introduced Tao's Law (τ-Law), a new guiding principle for semiconductor development. Its core concept centers on systematically reducing the time constant τ, using techniques such as logic folding to continuously compress signal propagation delay within chips, thereby increasing transistor density. Unlike traditional Moore's Law, which relies on shrinking transistors, Tao's Law provides an entirely new pathway for semiconductor evolution.

Huawei's semiconductor head He Tingbo recently published the V2 version of the Tao's Law paper, adding detailed engineering data and a ten-year product roadmap, underscoring Huawei's unwavering commitment to its independent chip technology strategy.

— CLS.cn

---

📰 Microsoft Lays Off 4,800 Employees, Sells Four Xbox Studios in AI Pivot

Microsoft kicked off its new fiscal year by laying off approximately 4,800 employees, representing 2.1% of its global workforce. Most affected employees come from the commercial sales organization and Xbox gaming division, with Xbox alone seeing about 1,600 layoffs on the day and plans to cut roughly 20% of Xbox jobs by fiscal year-end. Microsoft also confirmed the sale of four Xbox studios: Double Fine, Compulsion Games, Ninja Theory, and Undead Labs.

In an internal memo, Microsoft Chief People Officer Amy Coleman attributed the cuts to a changing technology landscape and AI's profound impact on how work gets done. However, she clarified that "the roles eliminated today are not being replaced by AI," framing the move as resource reallocation for operating in the AI era.

Notably, Microsoft has redeployed more than 4,000 employees to new roles over the past year, and over 30% of eligible employees opted for a voluntary retirement program. Nevertheless, the scale of layoffs underscores the intense restructuring pain tech giants face during the AI transition.

— The Verge

---

📰 World First: Fully Autonomous AI Agent Executes Ransomware Attack Without Human Direction

Security researchers at Sysdig have documented the industry's first confirmed case of a ransomware operation conducted entirely by a large language model (LLM) agent. Dubbed JadePuffer, the attack gained initial access by exploiting a remote code execution vulnerability (CVE-2025-3248) in the open-source Langflow framework. The LLM-driven agent then autonomously performed reconnaissance, credential theft, lateral movement, persistence, privilege escalation, and container escape—all without human direction.

Astonishingly, the AI agent demonstrated real-time adaptive capabilities, retrying failed steps with refined parameters in as little as 31 seconds. It ultimately encrypted 1,342 service configuration items, planted a backdoor administrator account, and left a ransom note demanding Bitcoin payment. No human operator directed the campaign at any point.

This incident marks a significant escalation in the threat landscape. Traditional ransomware relies on human operators or scripted toolkits, but JadePuffer proves that off-the-shelf LLM agents can now chain known vulnerabilities into complete end-to-end extortion workflows, raising urgent new requirements for AI security defense.

— BleepingComputer

---

📰 Reddit Fights LLM-Generated Spam Using LLMs, Launching an AI-vs-AI Arms Race

As powerful large language models make spam generation effortless, Reddit is now deploying LLM technology to combat the very problem LLMs created. The platform faces an unprecedented flood of AI-generated content—spam posts, bot comments, and fake accounts spreading at an industrial scale.

Reddit's strategy reflects a dilemma facing all internet platforms in the AI era: when anyone can generate massive volumes of content at near-zero cost using LLMs, traditional content moderation simply cannot keep pace. Platforms are being forced to deploy AI-powered detection systems to identify and remove AI-generated spam, sparking an "AI versus AI" arms race.

This challenge extends far beyond Reddit. The entire internet ecosystem is grappling with the content authenticity crisis brought by LLMs, as major platforms accelerate deployment of AI detection tools. However, this also raises new debates about false positive rates and the impact on genuine users.

— TechCrunch

---

📰 Canadian Government Overhauls Procurement Rules, Integrating AI to Help Small Businesses

The Canadian federal government has unveiled a new Small Business Procurement Program, representing a major overhaul of government purchasing rules designed to make it easier for small and medium-sized tech companies to win government contracts. The reform standardizes contract processes and introduces AI technology to streamline and accelerate procurement approvals.

This initiative carries significant implications for Canada's tech startup ecosystem. Historically, cumbersome government tendering processes have deterred small businesses, with large corporations dominating government IT and tech services procurement. By lowering barriers to entry and reducing administrative burdens, the new rules could open a multibillion-dollar market for Canadian tech startups.

Analysts view this as both a boon for Canadian tech companies and tangible progress in the government's digital transformation. By embedding AI into procurement systems, the government not only improves efficiency but also demonstrates confidence in homegrown AI technology by leading through example.

— BetaKit

---

📰 Katalyze AI Builds Agentic Operating System for Pharmaceutical Operations

Montreal-based AI startup Katalyze AI is building an "agentic operating system" for pharmaceutical companies—a platform that allows pharma workers to assemble teams of AI agents to tackle engineering, scientific, and manufacturing tasks. The core concept involves orchestrating multiple specialized AI agents to collaborate on different stages of the complex drug development process.

Katalyze's approach reflects the industry's evolution from single-model AI applications toward multi-agent collaboration. The pharmaceutical industry has long struggled with lengthy R&D cycles and high costs. AI agent systems promise to dramatically improve efficiency by automating experimental design, data analysis, and regulatory compliance documentation.

Backed by Inovia Capital, this Montreal startup represents the deepening of Canada's AI ecosystem from foundational research toward vertical industry applications. As pharmaceutical companies increasingly embrace AI, Katalyze's multi-agent collaboration solution is poised to play an increasingly vital role in drug development workflows.

— BetaKit