South Korea Locks In a $950 Billion AI Chip Alliance as Nvidia's Jensen Huang Makes His X Debut — July 26, 2026

⚡ Top Story

South Korea Unveils a $950 Billion AI Chip Alliance — Samsung and SK Hynix Lock In Nvidia and Broadcom Megadeals

South Korea's presidential policy chief Kim Yong-bum announced on July 24–25 that Samsung Electronics and SK Group have sealed roughly $950 billion in new AI deals with US tech giants, unveiled at a high-level summit in San Francisco bringing together Samsung Chairman Lee Jae-yong, SK Group Chairman Chey Tae-won, and global tech leaders. SK Group alone signed deals worth $750 billion, anchored by an SK Hynix–Nvidia partnership valued at more than $500 billion to secure next-generation memory supply. Separately, Samsung Electronics signed an MOU with Broadcom covering up to $200 billion across memory chips, foundry services, and advanced packaging.

Why it matters: This is the largest single national AI-hardware commitment disclosed to date, and it cements South Korea's memory-chip industry as load-bearing infrastructure for the entire frontier-AI buildout — a bet that Samsung and SK Hynix, not just Nvidia and the hyperscalers, are now indispensable chokepoints in the AI supply chain.

Sources: Reuters via Investing.com: Samsung Elec, SK Group seal $950 billion deals as South Korea hosts AI powers · Gulf News: Samsung, SK hynix clinch $950 billion AI chip deals with Nvidia · Tech Times: Samsung and SK Hynix Lock In AI Chip Supply Through 2030

🔬 Research & Papers

Nothing independently verified as newly published within the strict last-24-hour window met the bar for inclusion. A sweep of arXiv (cs.AI, cs.LG, cs.CL, cs.CV) and major lab research pages turned up nothing dated July 25–26 that wasn't either preliminary or already covered.

🏢 Industry & Startups

Databricks and Microsoft Extend Their Partnership Into the 2030s ⚠️ Dated July 23 — outside the strict 24-hour window, but not covered in any prior briefing.

Microsoft and Databricks announced an expansion of their decade-long partnership on July 23, extending it into the 2030s. Databricks will deepen its use of Azure Databricks — including Azure Cobalt, Microsoft's Arm-based infrastructure — to run its own lakehouse, while Microsoft integrates Databricks' Genie AI co-worker more deeply across its products.

Why it matters: A long-horizon commitment between two companies that could plausibly compete (cloud data platform vs. hyperscaler) instead locking arms for another decade, signaling neither sees a near-term reason to unwind the dependency.

Source: Microsoft: Databricks and Microsoft expand partnership

🛠️ Tools & Releases

Moonshot AI Set to Drop Kimi K3's Full Open Weights — Largest Open-Weight Model Ever

Moonshot AI's Kimi K3, a 2.8-trillion-parameter sparse mixture-of-experts model first previewed July 16, is releasing its full open weights at 00:00 UTC on July 27 (the evening of July 26 in US time zones) under a modified MIT license. The weights total roughly 1.4 terabytes using MXFP4 quantization. Kimi K3 already tops the Frontend Code Arena leaderboard (1,679 points), ahead of Claude Fable 5 (1,631), GPT-5.6 Sol (1,618), and GLM-5.2 (1,587), and supports a 1-million-token context window with image and video understanding. API access is priced at $3/$15 per million input/output tokens.

Why it matters: As a sparse MoE, only a fraction of its 2.8T parameters activate per request — a concrete signal that Chinese labs are routing around US compute constraints with architecture rather than raw chip access, while still shipping a model competitive with closed frontier systems.

Sources: VentureBeat: China's Moonshot AI releases Kimi K3, the largest open-source model ever · Tom's Hardware: China's 2.8-trillion-parameter Kimi K3 beats Claude Fable 5 in Frontend Code Arena

🌏 Global AI & Geopolitics

HCLTech to Build ₹14,257 Crore (~$1.7B) AI Data Centre in Odisha, India

HCLTech announced on July 24 it will build its first AI data centre in Odisha, India, with a planned capital outlay of ₹14,257 crore (roughly $1.7 billion), positioning the state as an emerging data-centre and digital-infrastructure hub as India competes for a slice of global AI compute buildout.

Why it matters: A concrete, large-dollar data point in India's push to become an AI infrastructure location rather than purely a consumption market — arriving the same week South Korea's chip alliance underscores how much of the AI race is now being fought over hardware and siting, not models.

Source: Daily Current Affairs: HCLTech Odisha AI Data Centre

⚡ Energy, Infrastructure & Chips

Nothing new beyond South Korea's $950 billion chip alliance (see Top Story) and India's Odisha data-centre investment (see Global section) was independently verified in the strict 24-hour window.

🤖 AI Agents & Autonomy

Nothing independently verified as newly announced in the last 24 hours met the bar for inclusion.

🔒 Safety, Alignment & Ethics

Florida Man Sues OpenAI, Says ChatGPT's Medical Advice Nearly Killed Him ⚠️ First reported July 22–23 — outside the strict 24-hour window, but not covered in any prior briefing.

Scott Winters, a former pastor, filed suit against OpenAI and Sam Altman in San Francisco County Superior Court, alleging ChatGPT dismissed early symptoms of a serious medical issue, told him he'd need eight to ten more episodes before it would be "serious," and advised him to stay home and remain "recliner-bound." Weeks later he suffered a near-fatal pulmonary embolism, which a treating doctor linked to the immobility ChatGPT recommended. The suit alleges negligence and unauthorized practice of medicine, and is among the first to argue a general-purpose chatbot should carry liability for bad medical advice.

Why it matters: It's a concrete legal test of whether chatbot output can trigger medical-negligence-style liability — a meaningfully different theory than the chatbot-companion and defamation suits that have dominated AI litigation so far.

Sources: CBS News: ChatGPT's medical advice nearly killed a Florida man, lawsuit against OpenAI claims · Courthouse News: Man sues OpenAI over 'dangerous' medical advice from ChatGPT

📊 Numbers & Signals

$950 billion — South Korea's newly announced AI deal package (Samsung, SK Group, US tech firms)
$750 billion / $500 billion+ — SK Group's total new AI deals / the SK Hynix–Nvidia partnership within them
Up to $200 billion — Samsung Electronics' new MOU scope with Broadcom
50 — signatories on Nvidia CEO Jensen Huang's "Open Weights" letter to Congress, up from 25 in a single day
2.8 trillion — parameters in Moonshot AI's Kimi K3, the largest open-weight model released to date
~$1.7 billion (₹14,257 crore) — HCLTech's planned AI data centre investment in Odisha, India

🧠 Worth Thinking About

Three stories today, read together, point at the same shift: the AI race's center of gravity is moving from models to hardware and geography. South Korea just disclosed the largest national AI-chip commitment yet. India is courting AI data centres state-by-state. And a Chinese lab is about to release the largest open-weight model ever built, explicitly designed around sparse activation to route around the very chip constraints Korea's deal is meant to reinforce. None of this is about which model reasons better — it's about who controls the memory, the packaging, the power, and the silicon that every model, closed or open, ultimately runs on.

🏛️ Government & Regulation

Nvidia's Jensen Huang Uses His First-Ever X Post to Push Congress Against Restricting Open-Weight AI

Jensen Huang joined X for the first time on July 24 to share "Open Weights and American AI Leadership,” an open letter warning Washington not to restrict downloadable AI models the way some once feared restricting open-source software in the 1980s. The letter's signatory list — initially 25 companies including Nvidia, Microsoft, Meta, and Palantir — doubled to 50 within a single day, adding OpenAI and Google. Amazon and Anthropic remain notably absent.

Why it matters: It's a rapidly consolidating industry coalition, missing two major labs, pushing back against possible legislation restricting open-weight models — arriving the same week a Chinese lab (Moonshot, see Tools & Releases) is about to ship the largest open-weight model in history, sharpening the stakes of exactly what Congress might restrict.

Sources: Forbes: Huang's Open Weights Letter Doubled To 50 Without Amazon And Anthropic · Tom's Hardware: Nvidia and 24 other companies sign open-weights letter as Washington weighs Chinese AI model ban · Fortune: Jensen Huang just used his first ever X post to warn the AI industry

🔭 Frontier Lab Dispatch

Nvidia — July 24: CEO Jensen Huang made his first-ever X post to back an open letter urging Congress not to restrict open-weight AI models (see Government & Regulation).

Moonshot AI — July 26/27: Releasing the full open weights of Kimi K3, a 2.8-trillion-parameter model — the largest open-weight release in history (see Tools & Releases).

🔗 Quick Links

Tier 1 — Official / Primary Sources

Microsoft: Databricks and Microsoft expand partnership

Tier 3 — Tech & AI News Media