KIMI K3 MEANS MORE COMPUTE CONSTRAINS, IRAN CEASEFIRE BACK, BIG TECH EARNINGS | MARKET OPEN

SUMMARY

Amit Kukreja provides an in-depth analysis of current market dynamics, focusing on AI infrastructure, geopolitical developments in Iran, and the upcoming big tech earnings cycle. The discussion covers semiconductor volatility, data center deals, the impact of new AI models like Kimmy K3, and broader macroeconomic trends affecting investor sentiment.

MAIN POINTS

• Recent volatility in semiconductor stocks is discussed amid a rebound in AI infrastructure names following new data center deals and updates from Kimmy K3. [1:30]
• A deadly Iranian attack on a US air base in Jordan escalates tensions, with US officials and politicians responding forcefully and oil prices spiking. [7:00]
• Oil prices decline after reports of potential mediation and a proposed 10-day cessation of strikes between Iran and the US, though skepticism remains about the durability of such headlines. [13:01]
• Microsoft announces a major partnership with AMD to deploy Helios rack-scale systems for AI inference, signaling continued big tech investment in compute infrastructure. [19:00]
• Kimmy K3, a Chinese AI model, reports being compute-constrained due to overwhelming consumer demand, highlighting ongoing infrastructure bottlenecks. [21:00]
• Anthropic adjusts access to its Claude Fable 5 model in response to competition from Kimmy K3, indicating a race to the bottom in AI model pricing and accessibility. [24:00]
• Chinese AI models gain significant token share among US firms, raising competitiveness concerns and fueling demand for data center and semiconductor capacity. [28:00]
• Iran secures $2.8 billion in new customer contracts with leading AI developers, raising its 2026 ARR target and boosting its stock price. [31:00]
• Hut 8 Mining announces a $9.8 billion, 15-year AI data center lease, further underscoring the scale of recent data center deals in the sector. [34:00]
• Archer and Anduril unveil an autonomous attack VTOL aircraft, expanding Archer's ambitions beyond urban air mobility into defense. [41:02]
• The market opens with Google rising on rumors of a new AI chip, while Microsoft and semiconductor stocks react to recent partnership and earnings news. [44:02]
• Morgan Stanley reaffirms memory stocks as attractively priced, but highlights Broadcom and Nvidia as the best risk-reward picks in semiconductors. [52:01]
• US strategic petroleum reserves are at a 45-year low, increasing the urgency for diplomatic solutions in the Middle East amid ongoing oil market volatility. [56:00]
• Oracle announces a $16.5 billion AI data center in New Mexico, but faces regulatory and permitting challenges that could delay the project. [1:02:00]
• Geopolitical headlines, including attacks on commercial vessels and missile interceptions in the Middle East, contribute to market volatility. [1:09:00]
• Debate continues over whether the proliferation of open-source Chinese AI models will reduce big tech's capex needs or simply shift demand across the compute stack. [1:19:00]
• Market breadth increases as investors rotate into small caps, healthcare, financials, and industrials amid AI fatigue and semiconductor volatility. [1:22:00]
• Microsoft's AMD partnership and Google's rumored frozen v2 chip highlight a shift toward vertically integrated AI infrastructure, with potential regulatory scrutiny on Chinese AI models. [1:38:00]
• Industry experts argue that open-source AI models will pressure token prices but not fundamentally undermine demand for compute and hardware providers. [1:45:00]
• Speculation arises over LeBron James's next team and its potential marketing impact, while Nvidia and other AI infrastructure stocks are discussed in the context of upcoming earnings. [1:57:00]
• Jersey Mike's prepares for an IPO, prompting comparisons to other consumer brands as investors weigh growth potential outside the AI sector. [2:05:00]
• Ongoing Iran-US tensions, maritime attacks, and the complexity of negotiating a lasting ceasefire are analyzed with input from regional experts. [2:07:00]
• Cumulus AI, a newly public neocloud company, orders $124 million in Nvidia GPUs, reflecting continued investment in AI infrastructure despite stock declines. [2:17:02]
• BlackRock plans to finance a $12 billion AI-focused data center in Texas, signaling institutional interest in data center infrastructure. [2:30:00]
• Circle CEO discusses regulatory progress for stablecoins and the potential impact of the Clarity Act and Genius Act on digital asset adoption. [2:31:00]
• Gavin Baker shares a cautiously optimistic outlook on AI infrastructure, emphasizing the shift of margins from model providers to compute and neocloud players. [2:48:01]
• The day closes with a review of major capex announcements, ongoing regulatory and geopolitical developments, and anticipation for key earnings reports from Google and Tesla. [3:03:00]

DETAILED ANALYSIS

The session opens with an overview of heightened volatility in semiconductor and AI infrastructure stocks, driven by a combination of new data center deals, AI model developments, and geopolitical events. Last week saw significant declines in semiconductor names, particularly after the emergence of Kimmy K3, a Chinese AI model that claims near-parity with leading US models at a fraction of the cost. This development initially spooked markets, as investors feared it would undermine the capital expenditure (capex) cycle supporting US chipmakers and data center operators.

However, the narrative quickly evolved when Kimmy K3 itself reported being compute-constrained, unable to meet surging consumer demand due to limited GPU and infrastructure capacity. This revelation reinforced the view that, regardless of model efficiency, the underlying demand for compute and data center resources remains robust.

Major US tech companies continue to drive the capex story. Microsoft announced a significant partnership with AMD to deploy Helios rack-scale systems for AI inference, marking a strategic shift toward vertically integrated AI infrastructure. The deal, which involves AMD's latest GPUs and CPUs, is set to begin shipments in the second half of 2026 and underscores Microsoft's willingness to invest heavily in compute.

Google, meanwhile, is rumored to be developing a new chip, codenamed frozen v2, designed to run its Gemini AI models more efficiently by embedding model architecture directly into silicon. This approach could yield substantial gains in speed and power efficiency, though it would reduce flexibility. The move aligns with broader industry trends, as both Microsoft and Meta are also pursuing custom chip development to optimize AI workloads.

The competitive landscape is further complicated by the rapid rise of open-source Chinese AI models. Kimmy K3 and Alibaba's new Quen 3 model have demonstrated that high-quality AI can be delivered at much lower costs, prompting US firms to reassess their spending strategies. The token share of Chinese models used by US companies has surged from under 10% in January to nearly 60%, reflecting a significant shift in adoption patterns.

This trend is bullish for AI infrastructure providers, as it drives greater demand for compute resources, but it poses challenges for closed model providers like Anthropic and OpenAI. In response, Anthropic has expanded access to its Claude Fable 5 model and adjusted pricing, signaling a potential race to the bottom in AI model economics.

Industry experts and venture capitalists, such as Gavin Baker and Byron Deeter, argue that while open-source models will pressure token prices and margins at the model layer, they will not fundamentally diminish demand for hardware and data center capacity. Instead, the shift may simply redistribute value from model providers to infrastructure players. The consensus is that the proliferation of open-source models, especially those from China, will accelerate the need for distributed compute and data center buildout, benefiting semiconductor manufacturers and neocloud operators.

Notably, Nvidia remains the dominant supplier, with over 90% market share in AI chips, but AMD is gaining ground through new partnerships.

The market has also seen a flurry of large-scale data center deals. Iran secured $2.8 billion in new contracts with leading AI developers, raising its 2026 annual recurring revenue (ARR) target to over $4 billion. Hut 8 Mining announced a $9.8 billion, 15-year AI data center lease, and Oracle revealed plans for a $16.5 billion AI data center in New Mexico, though regulatory hurdles may delay the project.

BlackRock is preparing to finance a $12 billion AI-focused data center in Texas, highlighting growing institutional interest in the sector. These deals, along with others involving CleanSpark, Nebius, and SpaceX, underscore the scale and momentum of AI infrastructure investment.

Geopolitical developments remain a key source of market uncertainty. Over the weekend, a deadly Iranian attack on a US air base in Jordan escalated tensions, prompting strong responses from US officials and a spike in oil prices. Subsequent reports of mediation efforts, including a proposed 10-day cessation of strikes, temporarily eased market fears and brought oil prices down.

However, skepticism persists regarding the durability of such ceasefire headlines, as both sides have a history of violating agreements. Additional attacks on commercial vessels and missile interceptions in the Middle East continue to contribute to market volatility. The US strategic petroleum reserve is at its lowest level in 45 years, intensifying the urgency for diplomatic solutions and influencing broader macroeconomic sentiment.

On the earnings front, the market is entering a pivotal period, with major tech companies such as Google, Tesla, IBM, and ServiceNow set to report results. The performance and guidance from these companies will be critical in shaping investor expectations for the capex cycle and the broader technology trade. Recent earnings from consumer-facing companies like Domino's and AMC have been strong, suggesting resilience in consumer demand despite macroeconomic headwinds.

However, software names have experienced mixed performance, with some like Palantir and Shopify holding up while others, including Adobe and Salesforce, face pressure as capital rotates into semiconductors and infrastructure plays.

Macro trends also influence market dynamics. The US faces a delicate balance between managing inflation, energy prices, and strategic reserves. Gasoline prices have risen above $4 per gallon nationwide, and the limited supply in the strategic petroleum reserve constrains the government's ability to stabilize prices.

The Federal Reserve's stance on interest rates remains a source of volatility, with recent comments from officials suggesting that further rate hikes may be inevitable. This environment has made high-beta and leveraged trades more challenging, as evidenced by the sharp decline in assets under management for levered semiconductor ETFs.

In the digital asset space, Circle's CEO discussed regulatory progress for stablecoins, highlighting the significance of the Genius Act and the potential impact of the Clarity Act on institutional adoption. Circle's transition from trading-focused stablecoin issuance to broader payment and infrastructure services reflects the maturation of the digital asset market. The company is positioning itself as a regulated national trust bank, aiming to serve as a foundational layer for digital dollars in the global financial system.

Throughout the session, the interplay between technological innovation, regulatory developments, and geopolitical risks remains central to market sentiment. The proliferation of open-source AI models and the resulting compute constraints reinforce the importance of infrastructure investment, even as questions linger about the long-term profitability of model providers. The upcoming earnings cycle will provide further clarity on the sustainability of the capex trade and the resilience of the broader technology sector.

Meanwhile, ongoing geopolitical tensions and macroeconomic challenges ensure that volatility will remain elevated in the near term.

LINKS

• Amit Kukreja's Twitter/X profile for market commentary and updates.
• Amit's Deep Dives Substack for in-depth analysis and research articles.

KEYWORDS