Semiconductor Chip Wars — 2026-03-30
The dominant story of the day is TSMC's 2-nanometer capacity being fully booked through 2028, creating a rare opening for Samsung Foundry to capture overflow demand from AI-hungry chip designers. Across manufacturing, geopolitics, and markets, the overarching theme is a widening supply-demand gap at the leading edge — fuelled by AI infrastructure buildout — while export-control enforcement holes and tariff uncertainty continue to shadow the industry's geopolitical landscape.
Semiconductor Chip Wars — 2026-03-30
Top Stories
TSMC's 2nm Capacity Fully Booked Through 2028 — Samsung Steps Into the Breach
TSMC's 2-nanometer process node is entirely sold out through 2028, according to new reporting, leaving major chip designers scrambling for alternatives. Samsung Foundry is emerging as the primary beneficiary, with sources indicating the Korean giant is positioning aggressively to absorb capacity overflow from TSMC's constrained roadmap. The dynamic is reshaping foundry competition at the leading edge just as AI spending accelerates.

AI Demand Chokes TSMC, Opens Door for Samsung
Samsung Electronics is emerging as a potential beneficiary of a tightening supply crunch at TSMC, as surging demand for artificial intelligence chips pushes TSMC's most advanced nodes to capacity. Industry observers note that customers who cannot secure TSMC allocations are increasingly evaluating Samsung Foundry for advanced node production, marking a potential inflection point for Samsung's market-share ambitions.

Samsung Targets Silicon Photonics Mass Production by 2028
Samsung Electronics is planning to begin mass production of silicon photonics starting in 2028 in a move that could reshape global data-center networking and AI hardware integration. The company is offering clients an integrated platform spanning photonics, AI semiconductors, and advanced packaging — a direct challenge to TSMC's co-packaged optics roadmap and a sign that the foundry competition now extends beyond pure transistor density.

Manufacturing & Supply Chain
TSMC 3nm Also Under Pressure. TSMC's 3nm node remains severely constrained, with the foundry reportedly prioritising "long-term, loyal" customers over newer clients. The squeeze extends across both N3 and N2, meaning the window for Samsung and other alternative foundries is unusually wide.
Samsung Silicon Photonics Platform Details. Samsung's 2028 silicon photonics ambition is coupled with an integrated AI semiconductor packaging pitch aimed at hyperscale data-center operators. The initiative leverages Samsung's combined strengths in memory, logic, and advanced packaging to differentiate from pure-play foundries.
$1 Trillion Semiconductor Market on Track for 2026. The Semiconductor Industry Association reported that the global chip industry delivered $791.7 billion in revenue in 2025 — up 25.6% year-on-year — and is on track to exceed $1 trillion in 2026. AI accelerator demand is the primary growth engine, with data-processing silicon forecast to exceed half of total semiconductor revenue by year-end.
Geopolitics & Trade Policy
Chinese Military Universities Acquired Sanctioned Nvidia Chips Despite Export Controls. Public procurement documents reveal that Chinese universities conducting military research completed purchases of Super Micro servers loaded with sanctioned Nvidia AI chips in both 2025 and 2026, highlighting persistent enforcement gaps in U.S. export-control architecture. The revelations raise fresh questions about the effectiveness of licensing regimes and end-user verification procedures.

U.S. Tariff Threats on Taiwan Chip Imports Remain Elevated. Ongoing U.S. tariff pressure on AI chip imports from Taiwan — stemming from national-security designations imposed earlier in the year — continues to cloud the supply chain. Real-estate and tech-sector observers in Taipei note the uncertainty is already affecting investment planning in the island's tech corridors.
U.S.–Taiwan Semiconductor Trade Framework in Place but Fragile. The United States classified chip and critical-mineral imports as national-security risks and implemented a 25% global semiconductor tariff earlier this year, alongside a bilateral framework with Taiwan requiring Taiwanese investment in U.S. chip manufacturing. The arrangement reduces Taiwan's headline tariff exposure but leaves the underlying architecture subject to executive revision.
Market Moves & Earnings
Foundry Market Leadership Debate Sharpens. With TSMC's 2nm fully booked, market-share dynamics at the leading edge are in flux for the first time in years. Samsung Foundry's willingness to take on overflow customers at N2 and N3 equivalents is being closely watched by analysts as a potential inflection point in the long-running TSMC dominance narrative.
AI Chip Sector Heads Toward $300–$350 Billion by 2029–2030. Creative Strategies projects AI accelerators — which accounted for under $100 billion in 2024 — will reach $300 billion to $350 billion in annual revenue by 2029 or 2030. The trajectory reinforces why TSMC's advanced nodes are overwhelmingly allocated to AI workloads, squeezing traditional compute and mobile customers down the priority queue.
Deep Dive: TSMC's 2nm Blackout and the Samsung Opportunity
The confirmation that TSMC's 2-nanometer process is fully booked through 2028 is the most consequential foundry news of the year so far, and its ripple effects are still being mapped. For context: this is not a soft allocation preference — it means that chip designers approaching TSMC today for N2 capacity are being told there is simply no room, regardless of price or relationship history.
The immediate strategic implication is that Samsung Foundry — which has spent years trying to chip away at TSMC's near-monopoly on the most advanced nodes — suddenly has a window that market forces, not just engineering progress, have opened for it. Samsung's 3GAP and 2GAP gate-all-around processes are the natural destination for designers locked out of TSMC N2. The Korea Herald reports that customers including those evaluating AI accelerator tape-outs are actively looking at Samsung as a credible alternative.
This matters beyond a simple capacity story. For years, Samsung has struggled with yield perception problems at advanced nodes that made fabless customers reluctant to risk their most critical designs there. If TSMC's books are closed, customers may be willing to accept higher risk or invest engineering resources in yield-optimisation work at Samsung — effectively giving Samsung the design-win volume it needs to improve its processes organically.
For Intel Foundry, the timing is doubly significant. Intel has staked its turnaround on winning external foundry customers with its 18A process. TSMC's full books theoretically benefit Intel too — but Intel's 18A is not yet in high-volume production, and Samsung's capacity is available now. The race is whether Intel can reach HVM credibility before the current TSMC overflow finds a permanent home at Samsung. Chinese foundries, meanwhile, remain locked out of sub-7nm advanced nodes by equipment controls, meaning the overflow battle is effectively a two-horse race between Samsung and Intel.
What to Watch Next Week
- Samsung Foundry Customer Announcements: Watch for any confirmed design-win disclosures from AI chip companies routing tape-outs to Samsung N2/N3 nodes in response to TSMC's 2028 booking closure.
- Export-Control Enforcement Response: The Super Micro/Nvidia chip smuggling revelations will likely trigger congressional and BIS responses; watch for new end-user verification guidance or additional entity-list additions.
- TSMC Arizona Phase 2 Updates: TSMC's U.S. fab ramp timeline is directly relevant to domestic customers locked out of Taiwan N2 capacity — any update on Arizona N2 qualification timelines will be market-moving.
- Samsung Silicon Photonics Partner Announcements: The 2028 silicon photonics roadmap will require hyperscale design partnerships announced well in advance; early customer signals from AWS, Google, or Microsoft would validate the strategy.
This content was collected, curated, and summarized entirely by AI — including how and what to gather. It may contain inaccuracies. Crew does not guarantee the accuracy of any information presented here. Always verify facts on your own before acting on them. Crew assumes no legal liability for any consequences arising from reliance on this content.
Create your own signal
Describe what you want to know, and AI will curate it for you automatically.
Create Signal