Trump, Xi, Huang Forge Chip Pact.

Mazen Al Sabbagh

— 8 min read
Share

TL;DR

  • Geopolitical Chip Play: The U.S. approved Nvidia H200 sales to ten Chinese entities, including Alibaba and Tencent. Beijing, however, is deferring deliveries as Trump, Xi, and Jensen Huang convene in Beijing. Xi affirmed China's market "will only open wider" to U.S. executives.
  • Cerebras IPO Surge: Cerebras Systems (CBRS) debuted on Nasdaq at $350 per share, an 89% increase from its $185 IPO price. This marks the largest U.S. IPO of 2026, with mid-afternoon trading settling near $325.
  • Bilateral AI Protocol: Treasury Secretary Bessent announced the inaugural U.S.-China formal dialogue on AI safety, establishing a critical bilateral framework between the two nations.
  • Microsoft's Strategic Hedge: Microsoft is reportedly exploring AI startup acquisitions, including Cursor, signaling a strategic move toward reduced reliance on OpenAI following a $100 billion partnership investment.
  • Musk-Altman Trial Concludes: Closing arguments in Musk v. Altman are underway; Musk's counsel cited his client's absence and highlighted five witnesses who called Altman a "liar." An advisory verdict is anticipated next week.

Lead Story: Trump, Xi, Huang Forge Chip Pact

This week’s most significant geopolitical maneuver in AI is unfolding in Beijing. The U.S. government has granted clearance for approximately ten Chinese entities, including Alibaba, Tencent, and ByteDance, to procure Nvidia’s H200 AI chips, as Reuters exclusively reported. Yet, not a single unit has been delivered. Beijing has stalled the transfers, creating a critical operational impasse while Nvidia CEO Jensen Huang is present at the Trump-Xi summit.

Huang's presence was orchestrated by a last-minute personal appeal from Trump, with Huang joining Air Force One from Alaska, per the NYT and CNBC. The H200, Nvidia’s second-tier AI chip, signals a calculated concession rather than unfettered market access, illustrating the delicate balance of this negotiation.

Midday update: The summit yielded its first concrete AI policy outcome. Treasury Secretary Scott Bessent confirmed the establishment of formal U.S.-China AI safety dialogues, marking the first bilateral safety protocol between the nations, CNBC noted. Bessent underscored Washington's intent to instill "U.S. values" into global AI standards while preserving its technological edge. Concurrently, President Xi Jinping addressed seventeen U.S. CEOs, including Musk, Huang, and Tim Cook, asserting that China's market "will only open wider," CNBC reported. The CEOs reciprocated, emphasizing the strategic importance of the Chinese market.

Nvidia commanded nearly 95% of China's advanced chip market prior to tightened export controls. Chinese tech firms, CNBC observed, have been aggressively developing domestic alternatives, notably Huawei’s Ascend chips, precisely to mitigate risks associated with uncertain foreign access. As the summit concludes Friday, any joint communique on technology, chip policy, or the new AI safety framework will be closely scrutinized.

In Other News

Cerebras Debuts with Explosive IPO. Cerebras Systems successfully priced its IPO at $185 per share Wednesday evening, exceeding its projected $150-160 range and securing $5.55 billion, Reuters confirmed. Trading commenced today on Nasdaq under CBRS, with shares indicated to open nearly 90% above IPO price, approaching $350 and valuing the company at over $70 billion, as Reuters further detailed. The offering was reportedly 20x oversubscribed. Pre-IPO acquisition discussions with Arm and SoftBank were noted by Seeking Alpha, and Sam Altman’s personal stake in Cerebras became a conflict point in the Musk trial. By mid-afternoon, CBRS settled around $325, maintaining a strong ~75% gain from its IPO price. This 89% opening surge on a multi-billion dollar raise signals more than bullish sentiment for Cerebras; it validates the appetite for the entire AI IPO pipeline, with entities like SpaceX and Anthropic closely observing. Public markets are not merely receptive to AI hardware; they are voracious.

Microsoft Diversifies AI Investments, Eyeing Post-OpenAI Future. Microsoft is actively pursuing AI startup acquisitions, signaling a strategic pivot towards reduced dependency on OpenAI, Reuters exclusively reported. The company reportedly considered acquiring code-generation startup Cursor this spring. This comes as a Microsoft executive testified in the Musk v. Altman trial, revealing over $100 billion invested in the OpenAI partnership to date, Bloomberg noted — a figure underscoring the relationship's profound scale and inherent risk. Further, Seeking Alpha highlighted OpenAI's renegotiation of revenue-sharing terms, which cap Microsoft's payouts at $38 billion by 2030 and eliminate Azure exclusivity. Microsoft is strategically de-risking its largest AI position. The substantial investment, the exploration of targets like Cursor, and the revenue-sharing cap collectively indicate a shift: the OpenAI partnership is evolving from an expansive alliance to a meticulously managed dependency. This represents a significant, if gradual, re-architecture of the industry’s most critical AI relationship.

Musk v. Altman: Closing Arguments Underway, Advisory Verdict Pending. Witness testimony concluded Wednesday in the highly publicized Musk v. Altman trial. The final day’s proceedings featured the introduction of a "jackass" trophy as evidence, a reference to Reid Hoffman, and an outburst from Musk, Business Insider reported. Musk subsequently departed for Beijing mid-trial, placed on "recall status" by the judge, requiring his availability to return if OpenAI requests, CBS News noted. Bloomberg Law's review characterized the trial as defined by "spectacle and personalities." Closing arguments began Thursday, with Musk’s lead counsel, Steven Molo, offering an apology for his client’s 6,000-mile absence in Beijing, CNBC reported. Molo then spent two hours challenging Altman’s credibility, citing five witnesses who allegedly called the OpenAI CEO a "liar." OpenAI’s Sarah Eddy countered with a precise legal argument: "No one in this case, other than Elon Musk, has testified to any commitments or promises that Sam Altman or Greg Brockman or OpenAI made to Mr. Musk," per the NYT. A nine-person advisory jury is expected to deliberate next week, with Musk pursuing up to $150 billion in disgorgement and Altman’s removal. Irrespective of the advisory verdict, this trial has generated a public record detailing OpenAI’s internal dysfunctions, Altman’s intricate financial positions, Microsoft’s $100 billion exposure, and Musk’s documented contradictions. The reputational impact is already indelible.

Anduril Secures $5B Series H, Valued at $61B for AI Defense. Anduril, the defense technology startup, completed a $5 billion Series H funding round led by Thrive Capital and Andreessen Horowitz, effectively doubling its valuation to $61 billion, TechCrunch and Reuters reported. The company posted $2.2 billion in 2025 revenue and is developing space-based interceptors integral to Trump's $185 billion "Golden Dome" defense initiative. Anduril’s $61 billion valuation positions it alongside major enterprise tech players like Databricks and Stripe. This signals that the defense AI sector has transcended its niche status, now attracting premier investors and commanding valuations comparable to leading frontier model firms.

X / Social Pulse

H200 Deal & Summit Repercussions: The announcement of the Bessent AI safety talks generated polarized responses. Critics argued against negotiating safety protocols while China defers H200 deliveries, viewing it as a leverage play. Conversely, pragmatists underscored the inherent significance of establishing the first formal bilateral AI safety channel, irrespective of immediate outcomes. Xi's "open wider" address to U.S. CEOs largely elicited skepticism, with analysts recalling similar commitments made repeatedly since 2015 without significant shifts in market access.

Cerebras Market Entry: Cerebras's near-90% opening surge transformed its narrative from a successful pricing event into a market frenzy. The semiconductor sector is currently assessing whether this reflects authentic demand for AI hardware or a broader, speculative repricing of any entity associated with AI chip development. The previously noted acquisition discussions with Arm and SoftBank now appear significantly more complex given a $70 billion-plus implied valuation.

Microsoft-OpenAI Strategic Re-evaluation: Microsoft's pursuit of AI startups, coupled with the reported $100 billion OpenAI investment and the $38 billion revenue cap, became the central topic in enterprise AI discussions. Many commentators suggested this marks the beginning of a strategic divergence for the once unified OpenAI-Microsoft alliance. The specific interest in Cursor, an AI coding tool, highlights Microsoft’s potential direct competitive entry against OpenAI’s existing code generation capabilities.

Musk v. Altman Trial Conclusion: The closing day's defining moment was Musk's counsel apologizing for his client's absence while simultaneously presenting witness testimony alleging Altman's dishonesty. Legal commentators observed the irony: Musk's core argument relied on character, yet his own non-appearance diminished the gravity of his claims. OpenAI's Sarah Eddy delivered a clear retort: no witness apart from Musk himself corroborated any alleged binding commitments. While the advisory jury deliberates next week, the consensus maintains that reputational damage has already been sustained, independent of the legal outcome.

One to Watch

Medicare's ACCESS Model: The Stealth AI Catalyst. While immediate market attention gravitates toward chip geopolitics and IPO surges, CMS has quietly unveiled a decade-long Medicare payment model. This initiative establishes the inaugural federal reimbursement mechanism for AI agents facilitating patient monitoring between clinical consultations. TechCrunch noted this as a significant development largely unnoticed by the broader tech ecosystem. Launching July 5, 2026, the program links results-based reimbursement to critical outcomes like blood pressure regulation and chronic disease management. Concurrently, NBC News indicated that nearly two-thirds of U.S. physicians are already leveraging OpenEvidence, an AI-driven medical search tool, often unbeknownst to patients. This convergence of proactive regulatory frameworks and organic, widespread adoption signifies a profound shift in healthcare AI integration.

Quick Hits

  • Anthropic's Agent Reconfiguration: Claude subscriptions now incorporate tiered "Agent SDK" credits ($20-$200/month), concluding unconstrained compute arbitrage. Concurrently, Claude Code weekly limits are increased by 50% until July 13, a defensive measure against OpenAI's enterprise outreach via Codex, Axios notes.
  • Thinking Machines Lab Talent Drain: Mira Murati’s AI venture is experiencing significant researcher departures to Meta, OpenAI, and xAI, Business Insider reports, highlighting the challenge for well-capitalized startups competing with Big Tech's gravitational pull.
  • Baidu's Agent-Centric Vision: Robin Li, at Create 2026, articulated AI's transition from "model competition to the AI agent era," forecasting the emergence of "super individuals," as reported by TechNode.
  • Samsung Strike Update: Samsung offered to resume wage negotiations Thursday, Yonhap stated; the union dismissed the proposal as inadequate. A May 21 walkout remains probable.
  • Papal Condemnation of AI Warfare: Pope Leo XIV, speaking at La Sapienza University, cautioned that AI weaponry risks a "spiral of annihilation", advocating for enhanced oversight of both military and civilian AI development.

The pervasive theme today is the fluid nature of leverage—its acquisition, erosion, and associated costs. The U.S. approved H200 exports, yet Beijing retained deliveries. Cerebras, priced at $185, launched at $350. Microsoft is strategically de-risking a $100 billion partnership. Musk's counsel apologized for his client’s absence from the very trial he initiated. Power dynamics within AI are rapidly reconfiguring; the consequences of misjudging these shifts are becoming acutely apparent.

Immediate indicators to monitor before tomorrow's briefing include: (1) Cerebras (CBRS) closing performance, specifically whether its initial $350 valuation can be sustained; (2) any communique from the Trump-Xi summit's second day regarding chip exports, AI safety, or the newly established bilateral protocol; and (3) the timeline for the advisory jury’s deliberation in Musk v. Altman.

Looking ahead, key events include Nvidia's Q1 FY27 earnings on May 20, Google I/O on May 19, and the Samsung strike deadline of May 21. Anthropic’s substantial funding round, estimated at $30-50 billion with a $900-950 billion valuation, remains pending. Microsoft’s aggressive AI startup acquisition strategy has now emerged as a compelling, ongoing narrative.

Sources

Lock in. M. mazen@thorterminal.com

Read more