Daily News Coverage: Nvidia's $60B Coding Startup Deal, Broadcom Seeks AI Chip Financing with AI Companies, Anthropic Builds Custom Silicon Team
The CODEW Daily News Coverage | August 22, 2026
Good morning, people! NVIDIA agreed to pay $60 billion for Poolside AI model licensing and invested an additional $10 billion, valuing the startup at $120 billion. Broadcom is reportedly seeking over $60 billion in AI chip financing to support Anthropic and other AI companies. Anthropic hired a former Google TPU founding executive to lead its custom silicon efforts and launched a $35 million fund for open-source cyber defense. Meanwhile, Supermicro completed an independent investigation into its export compliance mechanisms, and a critical TrueConf Server vulnerability faces a CISA remediation deadline today. Here's today's briefing from The CODEW's The Newsroom.
THE LEAD
NVIDIA Agrees to Pay $60B for Poolside AI Model Licensing, Invests Additional $10B
What happened: NVIDIA has agreed to pay $60 billion to obtain a license for AI models from startup Poolside, while also extending a $10 billion investment that values the company at $120 billion pre-money. The company will also extend job offers to more than 100 Poolside employees. Poolside will continue to operate independently, with its three founders remaining in place. The startup, which has offices in the U.S. and Paris, develops models focused on automating computer programming, particularly for government and defense software.
Why it matters: The transaction follows a similar structure to NVIDIA's previous acquisition of Groq's assets, where it paid approximately $20 billion for non-exclusive licensing rights and absorbed most of the company's employees and founder. The deal has drawn criticism from some lawmakers who view such arrangements as a means to circumvent merger reviews. NVIDIA is aggressively investing in the AI ecosystem to accelerate technology adoption across industries, which it views as essential for sustaining long-term demand for its chips and systems.
Who is affected: Poolside's existing investors; AI model developers; competitors in the AI coding and government software space; regulators reviewing the transaction structure.
What to watch next: Whether the deal faces regulatory scrutiny and whether NVIDIA continues its pattern of licensing-and-hiring acquisitions to bypass traditional M&A reviews.
AI & Infrastructure
Anthropic Hires Google TPU Founding Executive to Lead Custom Silicon Push
Anthropic has hired Amir Salek, a former Google TPU project founding executive, to join its compute team, reporting to compute lead James Bradbury. Salek was responsible for delivering Google's first seven generations of TPU from 2013 to 2022 and previously worked at NVIDIA, where he founded and led its SoC business. Anthropic confirmed it is building a custom silicon team, with job postings explicitly stating the company is "deepening that investment by building a custom silicon team." The company has reportedly engaged Samsung Electronics to potentially manufacture the chip using Samsung's 2nm process and advanced packaging services.
Why it matters: The hire signals Anthropic's serious commitment to developing custom AI silicon, following the path of OpenAI (which has already developed the Jalapeño chip with Broadcom) and other major AI labs. Custom chips would help Anthropic address supply constraints and tailor designs to its specific workload requirements. The move comes as Anthropic prepares for a potential IPO and signs massive compute agreements with multiple providers — including Google, AWS, SpaceX, and AMD — totaling over 10 GW of capacity.
Anthropic Launches $35M Fund for Open-Source Cyber Defense, Claude Security Scan Now Available on Mythos 5
Anthropic has launched a $35 million fund to support open-source cyber defense initiatives. The company also announced that its Claude Security scanning functionality is now available on Claude Mythos 5.
Why it matters: As AI agents become increasingly capable of autonomous cyber operations, AI labs are investing in defensive capabilities. The $35 million fund positions Anthropic as a supporter of the open-source security community, potentially building goodwill and ecosystem relationships ahead of its IPO. The expansion of Claude Security scanning to Mythos 5 — Anthropic's most capable model — suggests the company is integrating security tooling directly into its flagship offering.
Semiconductors
Broadcom Reportedly Seeking Over $60B in AI Chip Financing to Support Anthropic.
Broadcom is reportedly in talks with multiple lenders for an AI chip financing transaction exceeding $60 billion. The financing would help AI companies, including Anthropic, secure chips and other critical AI infrastructure.
Why it matters: The massive financing follows Broadcom's record AI-driven quarters and its collaboration with OpenAI on the Jalapeño inference processor. The scale of the transaction — over $60 billion — underscores the capital intensity of the AI chip market and the growing demand for Broadcom's custom silicon capabilities. The financing would help Anthropic and other AI companies secure chip supply as competition for AI hardware intensifies.
Supermicro Completes Independent Investigation into Export Compliance Mechanisms
Server manufacturer Supermicro has completed an independent investigation into its export compliance mechanisms, according to reports.
Why it matters: The investigation comes amid heightened scrutiny of semiconductor and server companies' compliance with U.S. export controls, particularly regarding sales to China. Supermicro's completion of the investigation may signal a resolution to compliance concerns that could affect its ability to serve certain markets.
AMD CEO Assures CoWoS Capacity as Advanced Packaging Constraints Persist
AMD Chair and CEO Lisa Su stated that advanced packaging technology (EFB) is still in early development stages and that the company will ensure it has sufficient CoWoS capacity, adjusting capacity based on market demand.
Why it matters: CoWoS advanced packaging remains a critical bottleneck in the AI chip supply chain. AMD's assurance that it will maintain sufficient capacity signals confidence in its ability to meet AI chip demand. The comments come as TSMC's CoWoS capacity remains fully booked amid surging AI chip demand.
Cybersecurity
TrueConf Server Critical Vulnerability Faces CISA Remediation Deadline Today
A critical vulnerability in TrueConf Server (CVE-2026-72529) has been added to CISA's Known Exploited Vulnerabilities catalog, with a federal remediation deadline of August 22, 2026.
Why it matters: The August 22 deadline represents an active, urgent patch priority for federal agencies and organizations using TrueConf Server. TrueConf is widely used for video conferencing and collaboration in enterprise and government environments. Organizations should prioritize patching to avoid potential exploitation.
Daily Security Briefing: SilkParasite Espionage, 2,000 WordPress Sites Compromised, Copilot Flaws
Antiy's August 22 security briefing reported multiple major threats: A "SilkParasite" cyber-espionage campaign targeting Central Asian government agencies, assessed with medium confidence to have Chinese origins. Nearly 2,000 compromised WordPress sites are being used as malicious infrastructure in a "StopAndProtect" campaign affecting over 6,000 IP addresses. Microsoft Copilot Personal was found to have three vulnerabilities, collectively called "CoSnitch," including one that enables one-click data theft (CVE-2026-24301). Researchers also disclosed a remote Spectre side-channel attack against Cloudflare Workers that can leak sensitive data from other tenant Workers. A "Ransom Busters" group is posing as a recovery service to extort companies already hit by ransomware.
Why it matters: The breadth of threats — from state-sponsored espionage to IoT botnets to AI-powered vulnerability discovery — underscores the escalating and diverse nature of the cyber threat landscape. The Copilot vulnerabilities highlight the new attack surface created by AI assistants, while the Spectre attack on Cloudflare Workers demonstrates that even serverless platforms are vulnerable to side-channel attacks.
Markets
Tesla Surges 5% on Cybercab Launch Plans; Memory and Optical Stocks Outperform
Tesla shares surged 5.14% to $362.86 on news that its Cybercab — the company's first vehicle without a steering wheel or brake pedals — could launch in Austin as soon as August. Meanwhile, U.S.-listed Chinese stocks showed mixed performance, with NetEase up nearly 8% and Alibaba down over 7%. Memory and optical communication stocks continued to show strength.
Why it matters: Tesla's Cybercab launch represents a significant milestone for autonomous driving and robotaxi commercialization. The stock surge reflects investor optimism about the company's autonomous vehicle strategy. The continued strength in memory and optical stocks reflects sustained AI infrastructure investment.
What to Watch
- NVIDIA-Poolside deal regulatory scrutiny: Whether the $60 billion licensing-and-hiring deal faces regulatory challenges from lawmakers concerned about M&A circumvention.
- Anthropic custom silicon: How Anthropic's custom chip development progresses under the leadership of the former Google TPU executive.
- Broadcom AI financing: Whether Broadcom closes its $60B+ AI debt deal and what it signals about AI infrastructure financing.
- TrueConf Server vulnerability: Whether organizations meet the August 22 CISA remediation deadline for CVE-2026-72529.
- Semiconductor chip price increases: STMicroelectronics is set to complete its third price adjustment of the year on August 23, with Analog Devices following on September 13.
English News Sources
- Asianet Newsable — Nvidia Reportedly Strikes $7B Licensing And Investment Deal With AI Startup Poolside (August 22, 2026)
- Seeking Alpha — Anthropic hires former Google chip developer as it pursues in-house solution (August 22, 2026)
- Bloomberg — Broadcom Seeks More Than $60 Billion in Latest AI Debt Deal (August 20, 2026)
- Yahoo Finance — Broadcom's (AVGO) Massive New Debt Deal Points To Where AI Is Headed (August 21, 2026)
- Bloomberg — Super Micro Says Top Management Didn't Know of Diversion Scheme (August 20, 2026)
- Yahoo Finance — SMCI Clears Senior Management In Export Control Investigation (August 22, 2026)
- BleepingComputer — CISA orders feds to patch actively exploited TrueConf Server flaws (August 20, 2026)
- Techmeme — Daily Tech News Aggregator (August 22, 2026)
- Edge n.tech — Anthropic hires Google chip veteran in $485K custom silicon push (August 22, 2026)
- Supermicro — Official Press Release: Completion of Independent Investigation (August 21, 2026)
Reviewed by Erwin Castro
on
Saturday, August 22, 2026
Rating: