Daily News Coverage: Anthropic Locks In $45B AI Compute Capacity Race, Amazon Confirms Purchase of 2M NVIDIA GPUs For AWS
The CODEW Daily News Coverage | August 27, 2026
Good Morning! Anthropic locked in $45 billion in AI compute capacity ahead of its IPO, while Nvidia reported blockbuster Q2 earnings with $96.2 billion in revenue and guided Q3 above $108 billion. Amazon committed to purchasing 2 million Nvidia GPUs, OpenAI's Jalapeño chip outperformed Nvidia's GB300 in key tests, and Salesforce beat earnings expectations with AI annual recurring revenue surpassing $3.9 billion. Meanwhile, a ransomware attack on a Philippine government system disrupted services for over 10 days. Here's today's briefing from The CODEW's The Newsroom.
THE LEAD
Anthropic Locks In $45B in AI Compute in Race to Secure Capacity Ahead of IPO
What happened: Anthropic has agreed to pay $45 billion over six years to Nscale, a UK-based AI infrastructure company, to secure approximately 460 megawatts of AI compute capacity at Nscale's Monarch data center in West Virginia. The facility, which will use Nvidia's latest Vera Rubin chips, is expected to begin operations by the end of 2027. The Monarch campus has a planned capacity of 1.35 gigawatts with a total investment of approximately $71 billion, including about $47 billion for AI chips. The project was originally planned for Microsoft, but Microsoft withdrew this summer, and Anthropic stepped in as the tenant.
Anthropic has been aggressively locking in compute capacity over the past eight months. The company has signed a $50 billion agreement with Fluidstack, a $10 billion deal with Volta Infra, and a $45 billion compute agreement with SpaceX — all as it prepares for a potential IPO. The company filed a confidential draft registration statement with the SEC in June and has been courting investors at a valuation of $965 billion.
Why it matters: The $45 billion Nscale deal — one of the largest AI infrastructure commitments in history — underscores the intensifying race for AI compute capacity ahead of the IPO window. Anthropic's aggressive compute buildout demonstrates that access to high-performance AI chips and data center capacity is the critical constraint for frontier AI labs. The deal also highlights the enormous capital requirements of AI infrastructure: Nscale's $71 billion Monarch campus investment alone exceeds the GDP of many countries. With Anthropic, OpenAI, Google, and Meta all competing for compute, the supply chain for chips, power, and data center capacity faces unprecedented demand.
Who is affected: Nvidia (primary chip supplier), Nscale and other AI infrastructure providers, hyperscalers competing for compute capacity, and IPO investors evaluating Anthropic's infrastructure commitments.
What to watch next: Whether Anthropic files its S-1 publicly in the coming weeks and how public markets value the company's infrastructure investments against its revenue growth.
AI & Infrastructure
Amazon Commits to Purchase 2 Million Nvidia GPUs as AI Infrastructure Spending Surges
Amazon and Nvidia announced that Amazon Web Services will purchase 2 million Nvidia GPUs and adopt Nvidia's new Vera CPU. Nvidia said AWS's Vera CPUs will be "integrated with Rubin" — Nvidia's upcoming AI chip — with some also used as standalone products. The deal demonstrates that hyperscalers continue to accelerate AI infrastructure investment despite concerns about spending moderation. Nvidia CFO Colette Kress said the company expects the "top five hyperscaler customers" to increase capital expenditures from $800 billion in 2026 to $1.3 trillion in 2027.
Why it matters: The 2 million GPU commitment — on top of Amazon's earlier pledge to install 1 million Nvidia chips this year — signals that hyperscaler AI infrastructure spending remains robust. The scale of the commitment ($100 billion-plus at current pricing) underscores Amazon's determination to remain competitive in the AI cloud market.
OpenAI's Jalapeño Chip Outperforms Nvidia GB300 in Key Test Metrics
OpenAI announced that its self-designed Jalapeño inference chip, developed in collaboration with Broadcom, outperformed Nvidia's GB300 on two key metrics: AI workload processing per unit of power and response speed. The chip is part of OpenAI's strategy to reduce dependence on Nvidia for AI infrastructure. OpenAI chip head Richard Ho confirmed the chip was compared against Nvidia's GB300 on the industry's leading public benchmark platform.
Why it matters: The successful test results mark a significant milestone in OpenAI's strategy to build its own AI infrastructure. OpenAI joins a growing list of AI companies — including Anthropic, Google, and Microsoft — that are developing custom silicon to control their own destiny in the AI chip market. The Jalapeño processor could give OpenAI a meaningful cost and performance advantage in inference workloads.
Semiconductors
Nvidia Q2 Revenue Hits $96.2B, Guides Q3 Above $108B as AI Infrastructure Demand Remains Strong.
Nvidia reported fiscal second-quarter revenue of $96.2 billion, more than doubling year-over-year and exceeding Wall Street expectations. The company guided third-quarter revenue to approximately $108 billion with a 2% margin of error, surpassing analyst expectations of $104.19 billion. Operating profit reached $63.7 billion. Nvidia now believes the commercial opportunity for its AI chips could exceed $1 trillion through 2027, doubling from the $500 billion opportunity the company previously cited.
Nvidia CFO Colette Kress said the top five hyperscaler customers are expected to increase capital expenditures from $800 billion in 2026 to $1.3 trillion in 2027, underscoring the sustained strength of AI infrastructure investment. Despite the strong results, Nvidia shares fell slightly in after-hours trading, down approximately 1%.
Why it matters: Nvidia's record results and $1 trillion opportunity estimate underscore that the AI infrastructure buildout is entering a new phase of scale. However, the after-hours selloff suggests investors are increasingly concerned about competitive pressures from custom silicon and the sustainability of hyperscaler capital expenditure growth.
Lam Research Breaks Ground on New Oregon Lab to Accelerate AI-Era Semiconductor R&D
Lam Research broke ground on a new research and development laboratory at its world-class R&D center in Tualatin, Oregon, to accelerate semiconductor innovation for the AI era. The new facility will focus on developing next-generation chip manufacturing technologies critical for AI processors.
Why it matters: The investment reflects the semiconductor industry's need to develop new manufacturing technologies to keep pace with AI chip demand. As chip architectures become more complex and power densities increase, advanced manufacturing processes become critical competitive differentiators.
Enterprise Software
Salesforce Q2 Beats Expectations, AI Products Surpass $3.9B Annual Recurring Revenue
Salesforce reported fiscal second-quarter revenue of $11.35 billion, up 11% year-over-year, exceeding analyst expectations of $11.33 billion. Adjusted earnings per share were $5.90, well above consensus estimates of $3.27. Net income surged 87% to $3.53 billion from $1.89 billion a year earlier. The company's AI products — Agentforce and Data Cloud — reached nearly $3.9 billion in annual recurring revenue, more than tripling year-over-year. Salesforce raised its full-year revenue guidance to $46.1–$46.4 billion and adjusted earnings per share to $16.67–$16.71. Shares surged 13% in after-hours trading to $231.93.
The company also expanded its partnership with Anthropic, launching "Claudeforce" — a plugin that combines Salesforce's business tools with Anthropic's Claude AI system. The offering includes more than three dozen prebuilt sales tools such as meeting preparation, deal health review, and pipeline review, and is available now to select pilot customers,s with an open beta planned for next month.
Salesforce CEO Marc Benioff called the quarterly results "one of the company's best reports to date" and cited AI as a key factor. "AI is delivering value across every layer of our platform. We're seeing incredible demand for our AI and data products," Benioff said. The company also announced it is acquiring customer service startup Fin for $3.6 billion and recorded a $2.6 billion investment gain from its stake in Anthropic.
Why it matters: Salesforce's results provide the clearest evidence yet that enterprise AI spending is translating into real revenue for software vendors. The $3.9 billion AI product ARR — more than tripling year-over-year — demonstrates that enterprises are willing to pay for agentic AI capabilities. The Anthropic partnership expansion signals that Salesforce is betting on AI as a growth driver and that the company's $21 billion investment in AI spending is paying off.
Cybersecurity
Philippine Government Seafarer System Hit by Ransomware, Services Disrupted for Over 10 Days
The Philippine Maritime Industry Authority's seafarer credentialing system was hit by a ransomware attack, causing nationwide service disruption for more than 10 days. Thousands of seafarers are unable to obtain the credentials required for overseas work, severely impacting their livelihoods. The agency is investigating the incident and working to restore services.
Why it matters: The attack highlights the vulnerability of government systems and the cascading economic impact of ransomware on critical public services. The Philippines is a major supplier of maritime labor — disruption to seafarer credentialing affects the global shipping industry and thousands of families dependent on overseas remittances.
KISA Launches Joint Cyber Exercise with 20 Asia-Pacific Nations on RMM Abuse Attacks
The Korea Internet & Security Agency (KISA) and the Asia-Pacific Computer Emergency Response Team (APCERT) launched a multinational joint cyber exercise on August 27 to simulate attacks exploiting Remote Monitoring and Management (RMM) tools. The exercise involves 20 Asia-Pacific countries and focuses on protecting enterprise internal networks.
Why it matters: RMM tools have become a favored initial access vector for ransomware groups, enabling attackers to move laterally across enterprise networks. The multinational exercise reflects growing international cooperation to address the escalating ransomware threat.
What to Watch
- Anthropic IPO filing: Whether Anthropic converts its confidential S-1 to a public filing in the coming weeks and how public markets value its $965 billion private valuation against its $45 billion compute commitments.
- Nvidia supply chain: Whether Nvidia can meet the surging demand from Amazon's 2 million GPU commitment and other hyperscaler orders without significant delays.
- OpenAI Jalapeño chip deployment: Whether OpenAI's custom chip moves from internal testing to production use and how it affects Nvidia's data center revenue.
- Enterprise AI spending: Whether Salesforce's $3.9 billion AI ARR growth accelerates or moderates through year-end.
- Hyperscaler capex: Whether the top five hyperscalers actually reach $1.3 trillion in capex in 2027 as Nvidia CFO Kress projects.
Source Attribution
- MarketScreener — Nvidia Beats Expectations Despite the Rise of In-House AI Chips (August 27, 2026)
- MarketScreener — Salesforce Posts Higher 2Q Profit, Expands Partnership with Anthropic (August 27, 2026)
Reviewed by Erwin Castro
on
Thursday, August 27, 2026
Rating: