1. Expansion of Long-Term Supply Agreements (LTAs) for Memory Drives Structural Price Increases in DRAM and NAND
• Core Source
“Samsung and SK Hynix are actively expanding long-term supply agreements (LTAs) in response to surging AI demand. While LTAs in the past served to smooth out price volatility, they are now evolving into a structure that actually drives prices higher.”
“As securing stable memory supply becomes essential for AI infrastructure expansion, major customers are signing long-term contracts while paying premiums to lock in production capacity. Meanwhile, those without contracts are being forced to pay ever-higher costs in an increasingly tight spot market.”
“SanDisk announced it has signed New Business Model (NBM) agreements with five customers. These contracts cover more than 33% of the company’s total bit supply in 2027, and include NAND supply obligations worth $42 billion, financial/cash guarantees of $11 billion, and (to date) $400 million in prepayments.”
“TrendForce raised its 2Q26 PC DRAM price forecast from the previous +40–45% to +43–48% (qoq).”
“Mobile DRAM prices are expected to surge +93–98% quarter-on-quarter in Q2, driven by suppliers’ efforts to close price gaps and customers’ competition to secure volumes.”
• Expected Impact
The fundamental nature of LTAs (long-term supply agreements) is undergoing a profound transformation. Previously functioning as a buffer to smooth out price swings from supply-demand imbalances, LTAs have now become a structural upward driver of memory prices amid the AI infrastructure investment boom. The root cause is a shift to a seller’s market. As Big Tech companies rush to lock in long-term contracts—even paying premiums—to secure HBM and high-capacity DRAM, those unable to sign contracts are forced to pay even higher prices in an increasingly tight spot market, creating a dual price escalation structure.
SanDisk’s NBM contract structure ($42 billion supply obligation, $11 billion financial/cash guarantee) is emblematic of this trend. In the NAND industry as well, long-term contracts have begun to function not merely as supply stabilization tools, but as price-setting mechanisms that restructure the industry itself. TrendForce’s upward revision of Q2 PC DRAM price forecasts to +43–48% (qoq), and the expected surge of +93–98% in mobile DRAM, reflects this structural shift.
For major memory suppliers like Samsung Electronics and SK Hynix, pricing power has reached historically high levels, and as the proportion of long-term contracts rises, the business is likely to evolve into a “low-volatility, high-margin” structure capable of defending profitability even during a potential downturn. Micron’s CEO stating that “memory is now a strategic asset” and forecasting that AI-related demand alone will absorb more than half of the entire DRAM and NAND market this year signals that this structural change is not a short-term phenomenon.
2. Anthropic Pursues $50 Billion Funding at $900 Billion Valuation and Establishes AI Services Joint Venture — Claude Ecosystem Expands Rapidly
• Core Source
“Reports indicate that Anthropic is pursuing its likely final pre-IPO funding round at a valuation of $900 billion. Given strong investor demand, a $1 trillion valuation is also being mentioned.”
“Anthropic’s annualized revenue (ARR) reportedly doubled in just two months to reach $44 billion. The core driver is the explosive growth of the programming agent Claude Code and surging enterprise customer demand.”
“The inference segment margin has risen from 38% to over 70%, significantly improving the quality of the business.”
“Blackstone, Hellman & Friedman, and Anthropic are core investors contributing $300 million each. Goldman Sachs is investing $150 million. Total fundraising including General Atlantic and others amounts to $1.5 billion.”
“The JV will serve as Anthropic’s consulting arm, supporting enterprise AI adoption including among private equity portfolio companies.”
• Expected Impact
Anthropic’s latest moves signal far more than a simple capital raise — they indicate that the monetization model of frontier AI companies is genuinely entering a mature phase. ARR doubling in just two months to reach $44 billion, combined with inference margins surging from 38% to over 70%, directly refutes the market concern that “AI model companies spend enormous capital with no path to profitability.” Notably, the fact that a single software development agent — Claude Code — is driving this growth means that AI has begun generating real, measurable ROI in workplace automation.
The establishment of the JV represents a strategic shift in Anthropic’s go-to-market approach, moving from API-centric sales to a deep enterprise integration model. The participation of major PE firms like Blackstone, Hellman & Friedman, and Goldman Sachs signals an intent to embed Claude into their vast portfolio companies — a structure analogous to Microsoft’s strategy of integrating OpenAI into the Office ecosystem. Reports that OpenAI is also establishing a $10 billion AI deployment JV with private equity firms underline the fact that competition for enterprise AI market leadership is rapidly shifting from “model performance” to “deployment and integration capabilities.”
From a valuation perspective, Anthropic’s implied value has more than doubled in just a few months — from $380 billion in the February Series G to $900 billion — reaching a level that could surpass OpenAI. This reflects the market’s recognition that the Claude-powered agent market (Claude Code, enterprise) has established itself as an independently high-growth market within the AI industry.
3. Hyperscaler Cloud Backlog Surpasses $2 Trillion — Confirming the Conversion of AI Infrastructure Demand into Real Committed Orders
• Core Source
“Google Cloud Backlog: $462 billion (+400% YoY, +$369.6 billion YoY, +93% QoQ, +$222.2 billion QoQ)”
“AWS Backlog: $364 billion (+93% YoY, +$175 billion YoY, +49% QoQ, +$120 billion QoQ)”
“When Microsoft and Oracle backlogs are added, the combined total exceeds $2 trillion (+176% YoY).”
“Growth began accelerating sharply from Q4 2025, when the diffusion of agents commenced, and the rate of increase further accelerated in Q1 of this year.”
“Google Cloud’s order backlog is currently at 23 times the level of a single quarter’s revenue.”
• Expected Impact
Cloud backlog (order intake) represents confirmed contract values not yet recognized as revenue. A rapid surge in this figure is therefore the strongest possible evidence that AI demand is not driven by market optimism or sentiment, but is underpinned by actual signed contracts — real demand in the truest sense. Google Cloud’s backlog standing at +400% year-on-year and 23 times a single quarter’s revenue means that strong revenue growth is already locked in for multiple quarters ahead.
The expansion of Agentic AI is the core driver of this surge. As AI in the form of autonomous agents — rather than simple chatbots — begins to be deployed across enterprise workflows, token consumption is growing exponentially, translating into a massive surge in cloud computing demand. Google’s management commenting that CapEx in 2027 will be “significantly higher” than 2026 due to “unprecedented internal and external demand for AI computing resources” speaks to the same dynamic.
For supply chain companies, this backlog surge signals that demand is not a short-term event. AWS’s $100 billion OpenAI computing contract, Google’s initiation of external TPU sales, and Microsoft’s $190 billion 2026 CapEx all represent demand that will convert into actual infrastructure procurement. This is precisely why structural, long-term demand for core supply chain players — Nvidia, TSMC, SK Hynix, Samsung Electronics — is virtually certain to persist.
4. Nvidia Accelerates CPO (Co-Packaged Optics) Commercialization by Five Years, Targeting Feynman GPUs in 2028
• Core Source
“Nvidia has moved up its CPO (Co-Packaged Optics) commercialization timeline from the previous 2033 to 2028 — a five-year acceleration.”
“CPO adoption is planned starting with the next-generation Feynman GPU.”
“Along with Feynman in 2028, Nvidia plans to deploy BlueField-5, NVLink 8 CPO, Spectrum 7 204T, and other AI platform chipsets.”
“With next-generation AI chips Rubin and Rubin Ultra in 2026, ABF substrate sizes are continuing to expand, with layer counts increasing to the 18–20 layer range.”
“The CPO market is projected to grow at a CAGR of 142% from 2026 to 2030.”
• Expected Impact
CPO (Co-Packaged Optics) replaces conventional copper cable-based electrical signal transmission with optical signals, enabling higher data transfer speeds while simultaneously reducing latency, power consumption, and heat generation. As AI data centers expand massively, the volume of data moving between servers and racks has exploded, and the physical limitations of copper-based transmission have begun to create bottlenecks. Nvidia’s decision to accelerate this timeline from 2033 to 2028 is a self-admission that the pace of AI infrastructure expansion has far exceeded initial projections.
Once CPO adoption begins in earnest with Feynman GPUs, a new wave of large-scale demand will form across related supply chains — including optical transceivers, silicon photonics, and test equipment. The CPO market’s projected CAGR of 142% from 2026 to 2030 does not represent a simple technology swap, but rather the birth of an entirely new market. AMD has also been mentioned as a potential CPO adopter for its MI500 GPU around 2028, suggesting this is not an Nvidia-only trend but a structural transformation sweeping the entire industry.
On the Korean front, Samsung Electro-Mechanics is expanding its FCBGA substrate business related to CPO optical interconnects, and Samsung Foundry has reportedly secured orders in the silicon photonics space, indicating that Korean supply chain companies are actively integrating into the CPO ecosystem. Taiwan’s ABF substrate makers such as Unimicron are also set to benefit from the high-specification demands accompanying increasing AI server substrate layer counts (18–20 layers).
5. AI Software Companies Shatter “AI Replacement” Fears and Prove Platform Value — Atlassian and Reddit Post Shocking Beats
• Core Source
“Atlassian (+29.58%) released earnings that alleviated fears that generative AI would threaten the viability of SaaS companies.”
“Atlassian reported that cloud revenue surged sharply and that customers are actually converging on its platform to leverage AI features, while long-term contracts are also expanding.”
“Ultimately, Atlassian’s results demonstrated that AI is not a force that renders software obsolete, but can act as a catalyst that enhances the value of existing platforms. Morgan Stanley and other major investment banks analyzed this growth as indicating that the pessimism that had weighed on the software sector was in error.”
“Reddit (+13.07%) surged sharply after reporting earnings that beat expectations on the back of a rapid surge in advertising revenue.”
“The market had previously expected that AI models like Anthropic’s Claude and GPT would directly perform workflows and coding, thereby replacing existing software tools, causing related stocks to fall persistently throughout the year.”
• Expected Impact
The biggest concern weighing on the software sector since 2025 was the fear that “AI models would directly replace existing SaaS tools.” The logic was that if Claude or GPT could directly perform coding, document writing, and workflows, collaboration tools like Atlassian’s Jira and Confluence would become unnecessary. However, Atlassian’s Q1 earnings proved this thesis wrong with actual data. Rather, enterprises are converging on the Atlassian platform precisely to leverage AI features, and this is confirmed by a surge in cloud revenue and expanding long-term contracts.
This is the first time the market has confirmed via earnings data that the relationship between AI and software platforms is one of “complementarity and augmentation” — not “replacement.” The more AI enhances productivity within a platform, the stronger the user lock-in effect becomes, and the greater the willingness to sign long-term contracts. Reddit’s case similarly refuted the fear that “AI proliferation would erode platform traffic and advertising revenue” with a sharp surge in ad revenue.
The simultaneous rally in major software companies including Salesforce, ServiceNow, and Intuit (up +4.13%, +3.23%, and +2.71% respectively) shows that Atlassian’s earnings served as a catalyst to change the narrative across the entire sector. This can be interpreted as a valuation re-rating opportunity for a software sector that had been excessively discounted due to AI threat narratives, with particularly positive re-rating expected for companies possessing strong platform moats and high customer switching costs.
Comment [0]