Tekin Morning | Wednesday, August 12, 2026
Good morning! Today we bring you six breaking tech stories covering cybersecurity, artificial intelligence, and aerospace.
Quick Morning Briefing
🎮
400 Microsoft Patches
- Record Patch Tuesday fixing North Korean Zero-Day
🎧
ChatGPT on Linux
- OpenAI launches native Linux desktop application
🚀
NVIDIA's $500B Deal
- Wall Street partners for AI compute infrastructure
🗡️
Gemini Hits 1B Users
- Fastest growing Google product in company history
📰
Windows 10 ESU Pricing
- $61 Year 1 Extended Security Updates announced
⚔️
SpaceX 93rd Launch
- 29 Starlink satellites delivered to low Earth orbit
Microsoft Breaks Records with 400+ Security Patches; North Korean Zero-Day Active in the Wild
Tuesday morning, August 13, 2026, Microsoft released its monthly Patch Tuesday update addressing over 400 security vulnerabilities across Windows, Office, Exchange, and other products in one massive drop. This isn't just a record number—it represents a fundamental shift in Redmond's security strategy. Among these, three zero-day vulnerabilities were identified, with one actively exploited in real-world attacks.
The primary vulnerability, CVE-2026-68820, resides in the Windows WinSock driver with a CVSS score of 7.0. Reports from Bleeping Computer and Krebs on Security indicate that this bug has been exploited by the Lazarus Group, linked to North Korea, in several targeted attacks. Microsoft's official statement confirmed that the MSRC rapid response team patched all Azure servers within 48 hours.
The remaining vulnerabilities span a wide range of products: Windows 10 and 11, Office, Exchange Server, SharePoint, Azure Active Directory, and Visual Studio. In terms of severity, 42 were classified as Critical, with most involving Remote Code Execution (RCE) and Privilege Escalation.
This marks the first time Microsoft has released over 400 CVEs in a single month. In 2025, the previous monthly record stood at 289. This surge indicates that internal security teams have intensified focus on Legacy Code, particularly following the 2024 security incidents and pressure from the U.S. Congress.
SecurityWeek analysts note that this volume signals a shift from "gradual release" to "mass release" strategy. Microsoft previously delayed lower-risk vulnerabilities to subsequent months, but now appears committed to releasing all identified issues at once to prevent security debt accumulation.
For organizations and IT administrators, this means a busy weekend ahead. Prioritization is crucial for determining which servers and clients to patch first. Microsoft recommends Exchange servers and Domain Controllers as top priority, followed by Windows 10 and 11 clients. For Windows 10, which transitioned to ESU (Extended Security Updates) on October 14, 2025, this update won't deploy automatically.
Another notable aspect involves two Publicly Disclosed vulnerabilities whose technical details were published online before patch release. These bugs in Office 2016 and SharePoint Server 2019 could allow attackers to execute malicious files via email or phishing links.
October 2003: Microsoft launches first Patch Tuesday
April 2017: Previous largest patch with 142 CVEs
May 2025: Record 289 vulnerabilities in one month
August 2026: New record with 400+ vulnerabilities
OpenAI Launches ChatGPT Desktop App for Linux; Now Covers Three Major Operating Systems
On August 11, 2026, OpenAI released the Preview version of its ChatGPT desktop application for major Linux distributions. The app includes three components: standard ChatGPT, ChatGPT Work for enterprises, and Codex, OpenAI's coding engine.
Linux distribution support is limited to Ubuntu 24.04 LTS, Ubuntu 26.04 LTS, Debian 13, Fedora 43, and Fedora 44. This selection indicates OpenAI's focus on distributions commonly used in software development and server environments.
TechCrunch reported that the app includes project management capabilities, browser integration, and native OS connectivity. Users can send commands directly from the terminal to ChatGPT or copy text via Clipboard for faster queries.
"
With the Linux release, OpenAI has covered all major operating systems. This shows they're more serious than ever about replacing GitHub Copilot.
Kyle Wiggers, TechCrunch
The timing is intriguing. OpenAI announced this simultaneously with Microsoft Build 2026, just as Microsoft was unveiling GitHub Copilot Workspace. This could signal direct competition between the two giants in AI-assisted coding.
Technically, the application is built on Electron with an install size of approximately 280MB. Some users on Reddit have complained this is excessive for a chat application, but OpenAI explains that most of this size consists of the Codex module and offline code prediction models.
A unique feature is offline Auto-Completion capability. Even with internet disconnected, Codex can provide code suggestions based on local models. However, this functionality is only enabled for popular languages like Python, JavaScript, TypeScript, and Go.
Comparing this app with competitors shows OpenAI's attempt to solidify its position in the Developer Tools market. GitHub Copilot, owned by Microsoft, still holds the largest market share but has limitations: it only works within VS Code and a few other IDEs, and costs $10 monthly.
ChatGPT Desktop has advantages in being IDE-independent and usable in any development environment. Additionally, ChatGPT Plus subscription ($20/month) includes all features including GPT-4, DALL-E 3, and Codex, while Copilot is designed solely for coding.
NVIDIA Mobilizes $500 Billion with Wall Street for AI Infrastructure Buildout
On August 10, 2026, NVIDIA announced strategic partnerships with six major American financial institutions (Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR) to establish an independent financing platform for AI compute infrastructure. The platform aims to mobilize over $500 billion in third-party capital over the coming years.
This deal signifies that NVIDIA is no longer just a chip maker but aspires to become an AI infrastructure architect. According to CNBC, this business model mirrors what occurs in commercial real estate or toll roads: you build infrastructure, rent it out, and generate stable recurring revenue.
Jensen Huang, NVIDIA's CEO, explained in an interview with CNBC's Becky Quick that this platform allows startups and mid-sized companies to access compute infrastructure without massive upfront costs. For instance, if a startup needs 1,000 H100 GPUs, they don't need to pay $30 million cash upfront. They can rent through this platform and pay monthly.
NVIDIA's move demonstrates that the AI war is shifting from chip sales to complete Turnkey Infrastructure provision. Now NVIDIA wants to operate like AWS or Azure, but with exclusive focus on AI workloads. This could pose a serious threat to Azure Machine Learning and Google Cloud TPU.
Larry Fink, BlackRock CEO, announced at the same event that this platform enables institutional investors to invest in AI infrastructure exactly like investing in office buildings or airports. He stated: "We believe AI Compute is a new Asset Class and will deliver stable returns of 8 to 12 percent annually."
The agreement includes building dedicated datacenters in the United States, Europe, and the Middle East. The first datacenter in Texas is under construction and will launch by end of 2026. Its initial capacity will be 500 megawatts of power, capable of powering over 100,000 H200 GPUs.
Interestingly, this platform is only available for Qualified Customers. Companies seeking to use this service must have at least $10 million market value and an AI-first business model. This restriction prevents misuse for Crypto Mining and other non-productive workloads.
🎧
Tekin Editorial Team
Editor's Note
NVIDIA's deal could mark an inflection point in tech industry history. For the first time, AI infrastructure is recognized as an independent Asset Class. This means investors can now invest directly in physical AI infrastructure instead of just buying tech stocks. If this model succeeds, we might witness a return to massive capital investments similar to the 1990s internet infrastructure boom.
Google Gemini Hits 1 Billion Users; Becomes Google's Fastest-Growing Product
On August 11, 2026, Sundar Pichai, Google CEO, announced on Twitter that the Gemini app has reached one billion monthly active users. This milestone makes Gemini the fastest product in Google history to reach this number. It's the 14th Google product to achieve this level, following Search, Gmail, Maps, YouTube, Chrome, Play
Store, and Android.
- One billion monthly active users in under 18 months
- 150 million images generated daily
- 63% of users utilize Voice feature
- 14th Google product to reach this milestone
Forbes reported that Gemini's growth stems from three main factors: First, integration with Android. In Android 15, Gemini replaced Google Assistant as the default assistant. Second, bundling with Chrome and Workspace. Every Gmail or Google Docs user automatically has Gemini access. Third, massive marketing campaign in India partnering with Jio, which has 505 million subscribers.
However, TechCrunch reminds us of an important caveat: user count doesn't necessarily equal success. A substantial portion of this one billion consists of people using Gemini due to OS integration, not conscious choice. This parallels what happened with Microsoft Edge: high user numbers but low active usage rate.
Nevertheless, Google claims Gemini's weekly Retention rate is increasing. In Q2 2026, 67% of users who installed Gemini used it at least once weekly. This number was only 42% in Q1.
One new capability added to Gemini this week is Imagen 3, an advanced image generation model capable of producing photorealistic images at 2048×2048 quality. This model is currently available only for Gemini Advanced users ($19.99/month), but Google promises it will become free for everyone by end of September.
| Product | Time to 1 Billion | Launch Year |
|---|
| Gemini | 18 months | 2024 |
| Chrome | 8 years | 2008 |
| Gmail | 12 years | 2004 |
| Maps | 10 years | 2005 |
| YouTube | 8 years | 2005 |
Windows 10 Reaches End of Road; ESU Pricing Structure Announced
Microsoft officially announced the pricing structure for the Extended Security Updates (ESU) program for Windows 10. According to this program, each device costs $61 for Year 1 (ending October 13, 2026), and this amount doubles each subsequent year until the three-year program concludes on January 8, 2030.
This means an organization with 1,000 Windows 10 devices must pay $61,000 for Year 1, $122,000 for Year 2, and $244,000 for Year 3. Over three years, each device will cost $427 total.
The Register reported that this pricing is deliberately expensive to push organizations toward Windows 11. Microsoft employed this same strategy for Windows 7 and successfully migrated 85% of enterprise users to Windows 10.
Microsoft offers three methods for receiving one year of free ESU: 1) Enable Windows Backup for settings sync, 2) Redeem 1,000 Microsoft Rewards points, 3) Purchase one-year license for $30. These options are available only for home users, not organizations.
Interestingly, Microsoft announced in June 2026 that home users could receive an additional free year of ESU. Tom's Hardware reported that users already enrolled in the program would automatically receive this extra year.
This decision followed strong community backlash. Many home users with older devices (e.g., without TPM 2.0 or seventh-gen Intel CPUs) cannot migrate to Windows 11 and are forced to continue using Windows 10.
Technically, the ESU program only covers Critical and Important security updates, not Feature Updates. This means no new capabilities will be added to Windows 10—only security holes will be patched.
Another ESU limitation is cumulative licensing. If an organization decides to enter the program in Year 2, they must also pay for Year 1. This rule prevents gaming the system and deliberate organizational delays.
PROS
- 3 free options for home users
- Covers Critical and Important security updates
- Annual renewal option available
CONS
- Extremely high price for organizations ($427 over 3 years)
- No Feature Update support
- Mandatory cumulative licensing
SpaceX Launches 29 More Starlink Satellites; 93rd Launch of 2026
Tuesday morning, August 11, SpaceX launched the Starlink 10-19 mission from Cape Canaveral Space Force Station in Florida. The Falcon 9 rocket lifted off at 11:58 a.m. EDT from Space Launch Complex 40, delivering 29 Starlink V2 Mini satellites to low Earth orbit.
This marked the 93rd Falcon 9 launch of 2026 and the 72nd dedicated Starlink mission. The first-stage booster successfully landed on the droneship in the Atlantic Ocean. This particular booster (number B1085) flew for its fifth time.
Space.com reported that this booster previously completed two astronaut missions: Crew-9 to the International Space Station in September 2024 and Fram2, which was Earth's first polar flight in March 2025. Using the same booster for different payloads demonstrates SpaceX's confidence in reusable booster reliability.
To date, SpaceX has conducted 93 Falcon 9 launches, with 72 dedicated to Starlink. This translates to approximately one launch every 3 days. The previous record in 2025 stood at 81 launches.
V2 Mini satellites represent an improved second-generation Starlink with wider bandwidth capacity and lower latency. Each satellite weighs approximately 800kg and can transfer 4 times more data than first-generation units.
One interesting aspect of this launch was the northeasterly trajectory. Typically, Starlink launches from Cape Canaveral head east or southeast, but this time a northeasterly path was chosen to place satellites in a higher polar orbit. This orbit is essential for covering northern Canada and Scandinavia.
SpaceX currently operates over 6,200 active Starlink satellites in orbit, with an ultimate goal of 42,000 satellites. However, this target requires FCC and international organization approval, which remains under review.
Another notable point is intensifying competition in the satellite internet market. Amazon with Project Kuiper and China with Project GW are entering this market. Amazon plans to launch its first commercial satellites by end of 2026, and China is building 13,000 satellites for its national project.
Examining SpaceX economics reveals how this company has revolutionized the space industry with its unique business model. Previously, launching a satellite to low Earth orbit cost approximately $10,000 per kilogram. SpaceX has reduced this to under $1,500.
This cost reduction stems from three main factors: booster reusability, rocket mass production, and supply chain verticalization. SpaceX manufactures most rocket components internally, while traditional companies like Boeing and Lockheed Martin use hundreds of external suppliers.
Wednesday morning, August 12, proves that security and infrastructure are driving the next tech evolution. Microsoft's record patch drop alongside NVIDIA's half-trillion-dollar Wall Street alliance signals a monumental shift toward enterprise-grade AI infrastructure readiness.
Should I update all Windows systems immediately?
Yes, especially if you run Exchange servers or Domain Controllers. The CVE-2026-68820 vulnerability is currently being exploited in real attacks and must be patched urgently.
Does ChatGPT work on Arch Linux?
No, currently only Ubuntu 24.04/26.04 LTS, Debian 13, and Fedora 43/44 are supported. Arch users can use the web version or wait for unofficial AUR packages.
Why is Windows 10 ESU pricing so high?
This pricing is deliberately expensive to push organizations toward Windows 11. Microsoft used this same strategy for Windows 7 and successfully migrated 85% of users.
Does NVIDIA's Wall Street deal mean GPU prices will drop?
No, this deal is for datacenter infrastructure, not consumer products. Consumer graphics card prices will likely remain unchanged.
How can I get free Windows 10 ESU?
Three methods exist: 1) Enable Windows Backup, 2) Redeem 1,000 Microsoft Rewards points, 3) Purchase one-year license for $30. These options are for home users only.