Back to News
IndustryAI Understanding briefing

TSMC approves $1 billion in stakes for OpenAI and Anthropic

TSMC's board authorized up to $500 million each for equity stakes in OpenAI and Anthropic, marking a strategic pivot to invest directly in frontier AI labs to secure demand visibility and manage supply constraints.

4 min readRead the original reporting
Source-page capture accompanying TSMC approves $1 billion in stakes for OpenAI and Anthropic
Attributed reportingSource recorded
Publisher
scmp.com
Source type
Reporting by a news outlet — not a first-party document.

What we could not confirm independently: This claim is attributed to the named outlet. We did not verify it against a first-party document. (scmp.com)

ContextUnderstand this in 60 seconds

Key terms

Compute
The processing resources required to train and run models, often measured in FLOPS or GPU hours.

What happened

TSMC's board authorized up to $500 million to buy Anthropic common shares at its IPO price and up to $500 million for OpenAI preferred shares, alongside a $500 million strategic investment arm for the semiconductor ecosystem.

Taiwan Semiconductor Manufacturing Company (TSMC) announced that its board authorized up to US$500 million to purchase common shares in Anthropic at its initial public offering (IPO) price. Additionally, the board approved up to US$500 million to buy preferred shares in OpenAI.

The board also separately approved up to US$500 million to establish a dedicated strategic investment arm focused on the semiconductor ecosystem. This represents a notable pivot for TSMC, as CEO C.C. Wei had previously dismissed the idea of investing in end-market consumers of its silicon as recently as July.

According to an updated IPO filing, Anthropic stated that TSMC's subsidiary, TSMC Partners, would buy non-voting Class N shares in a US$26 billion private placement alongside investors including Google and Nvidia. This transaction is subject to the offering's completion and other conditions.

The investment comes as TSMC approves approximately US$40.9 billion for projects including advanced node manufacturing, packaging capacity, and fab construction. Goldman Sachs forecasts TSMC's annual capital expenditure will climb to US$85 billion next year and reach US$98 billion in 2028.

Source details: scmp.com

Why it matters

This move signals a significant shift in TSMC's strategy from pure contract manufacturing to direct equity participation in the AI value chain. By investing in the end-users of its chips, TSMC aims to gain better visibility into actual demand, mitigate overcapacity risks, and secure its position in a market where chip design and AI model development are increasingly intertwined. It also highlights the growing financial interdependence between chipmakers and AI labs.

TSMC's move to take equity stakes in OpenAI and Anthropic provides the chipmaker with direct financial ties to the frontier AI labs driving demand for its products. This strategy allows TSMC to look beyond its immediate fabless customers, such as Nvidia and AMD, to gauge true end-user demand amid an unprecedented capital spending spree across the industry.

Holding equity stakes could help TSMC manage supply bottlenecks by giving it a closer look at the actual needs of companies buying its products. Anthropic's IPO filing noted that TSMC's capacity constraints had at times delayed scheduled chip deliveries, highlighting the critical nature of this relationship.

The investment reflects a blurring line between chip designers and AI model makers. Anthropic's filings indicated that partnerships with Amazon and Nvidia allow it to influence future processor architecture, while OpenAI is collaborating with Broadcom to deploy custom accelerators. TSMC's investment positions it to remain central to these evolving hardware-software dynamics.

Citi analyst Laura Chen noted that TSMC's five-year forecasting framework factors in demand for AI models and emphasizes the financial resilience of established cloud providers. This approach is designed to help mitigate overcapacity risks during potential cyclical corrections in the AI market.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
Interactive Concept Check+10 Points
AI Models Explained Quiz

Which component of an AI application is the machine-learning model itself?

What to watch next

Monitor the completion of Anthropic's IPO and OpenAI's private placement to confirm the actual investment amounts. Watch for any disclosed joint design agreements or chip supply arrangements between TSMC and these AI labs, which could further blur the lines between hardware and software providers.

The completion of Anthropic's IPO and OpenAI's private placement will determine the final investment amounts and terms. The board statement did not disclose whether the equity commitments included specific chip supply arrangements or joint design agreements.

TSMC's ability to manage supply bottlenecks and meet the demands of AI labs will be a key indicator of the success of this strategic pivot. Any further delays in chip deliveries could impact the relationship between TSMC and its AI customers.

The broader implications of TSMC's investment in the semiconductor ecosystem, including the new strategic investment arm, may reveal further moves to secure its position in the AI supply chain. This could include additional investments in other AI-related companies or technologies.

Related guides & quizzes

Found this useful?