Taiwanese chip designer MediaTek is positioning itself as a rare neutral player in one of the semiconductor industry’s most consequential battles, embracing both Taiwan Semiconductor Manufacturing Co. (TSMC) and Intel’s advanced packaging technologies as demand for artificial intelligence chips accelerates worldwide.

The strategy becomes necessary as packaging technologies that connect multiple chips into a single high-performance system have become nearly as important as the chips themselves. As AI models grow larger and more computationally demanding, semiconductor companies are increasingly relying on advanced packaging to overcome physical limitations in chip design and boost performance.

Speaking to reporters in Taipei on Friday, MediaTek Senior Vice President Vince Hu said the company supports both TSMC’s CoWoS and Intel’s EMIB packaging technologies, giving customers flexibility in how their AI processors are built.

“We’re one of the few custom silicon providers that support both CoWoS and EMIB. We let our customers choose,” Hu said.

MediaTek has been growing ambitions beyond its traditional smartphone processor business. The company has emerged as a significant player in the rapidly expanding market for custom AI chips, where technology firms are designing proprietary semiconductors tailored to their specific workloads rather than relying solely on off-the-shelf products.

TSMC’s CoWoS, short for Chip-on-Wafer-on-Substrate, has become one of the most sought-after technologies in the AI supply chain. It is widely used in high-performance processors, including Nvidia’s AI accelerators, which power many of the world’s leading data centers.

Intel’s EMIB, or Embedded Multi-die Interconnect Bridge, represents a competing approach that allows different chip components to be connected within a package while maintaining high-speed data transfer and lower power consumption.

MediaTek’s willingness to support both technologies is strategically significant as it is emerging at a time when customers are increasingly concerned about supply-chain concentration. TSMC’s CoWoS capacity has been stretched by explosive AI demand, prompting many technology companies to explore alternatives.

According to two people familiar with the matter, Intel’s EMIB technology is being evaluated for custom AI processors that MediaTek is designing for Alphabet’s Google. MediaTek has not publicly identified Google as a customer and declined to comment on whether EMIB would be used in chips destined for the U.S. technology giant.

If adopted, it would mark another important validation of Intel’s efforts to establish itself as a serious competitor in the advanced packaging market, an area traditionally dominated by TSMC.

The developments come as competition intensifies across the custom AI chip sector. Major cloud providers, including Google, Microsoft, and Amazon, are designing their own processors to reduce dependence on Nvidia and optimize performance for specific AI workloads. That trend is creating a lucrative opportunity for companies such as MediaTek, which provide chip design expertise without requiring customers to build semiconductor teams from scratch.

Reflecting that opportunity, MediaTek reiterated that it has doubled its forecast for 2026 data-center revenue to $2 billion, signaling confidence that AI-related demand will remain strong.

The company estimates the global market for custom AI application-specific integrated circuits (ASICs) could reach between $70 billion and $80 billion by 2027. MediaTek is targeting a 10% to 15% share of that market, a goal that would establish it as one of the leading beneficiaries of the next phase of AI infrastructure spending.

The outlook aligns with broader industry expectations that custom silicon will account for an increasing share of AI workloads as hyperscalers seek greater efficiency and lower operating costs.

MediaTek is also preparing for the next generation of semiconductor manufacturing technology. The company disclosed that it already has multiple test chips running on TSMC’s A14 process, the Taiwanese foundry’s next-generation manufacturing node that is expected to enter mass production in 2028.

Early engagement with A14 suggests MediaTek intends to remain at the forefront of advanced chip development as customers pursue increasingly sophisticated AI systems requiring higher performance and improved power efficiency.

In another notable move, MediaTek said it plans to manufacture chips at TSMC’s Arizona facilities, including products built using 4-nanometre and 3-nanometre technologies.

The decision shows how companies are embracing the growing importance of geographic diversification in semiconductor manufacturing. Governments and technology companies have increasingly sought to reduce supply-chain risks by expanding production outside Asia, particularly amid rising geopolitical tensions involving China, Taiwan, and the United States.

Thus, supporting both Intel and TSMC packaging ecosystems while simultaneously expanding its manufacturing footprint into the United States provides MediaTek with flexibility at a time when AI demand is reshaping the semiconductor landscape.

As cloud providers race to deploy more powerful AI systems and reduce dependence on a handful of suppliers, companies capable of navigating multiple technology ecosystems are expected to gain a critical competitive advantage. MediaTek is betting that neutrality, flexibility, and deep engineering expertise will allow it to capture a larger share of one of the fastest-growing segments in the global technology industry.

OpenAI has granted select Japanese financial institutions early access to its GPT-5.5 artificial intelligence model, marking a significant step in the growing effort by governments and financial regulators to defend critical infrastructure against a new generation of AI-powered cyber threats.

Japan’s Finance Minister, Satsuki Katayama, announced the development on Friday after meeting OpenAI Chief Strategy Officer Jason Kwon in Tokyo, describing the move as “a big step forward in strengthening Japanese financial institutions’ ability to defend against cyberattacks.”

The decision comes as financial regulators worldwide grapple with the double-edged nature of rapidly advancing AI systems. While the latest models can help organizations identify software vulnerabilities, automate threat detection, and strengthen cyber defenses, they can also equip malicious actors with powerful new tools for discovering weaknesses, writing sophisticated attack code, and scaling cyber operations at unprecedented speed.

The emergence of highly capable coding-focused AI models has intensified concerns across the banking industry, where cybersecurity failures can have systemic consequences. Financial institutions are viewed as one of the sectors most exposed to AI-enabled attacks because of their interconnected networks, sensitive customer data and central role in the global economy.

According to Japan’s Nikkei newspaper, the institutions expected to receive access include the country’s three largest banking groups: Mitsubishi UFJ Financial Group, Sumitomo Mitsui Banking Corporation, and Mizuho Bank. Although the banks have not publicly disclosed how they intend to deploy the technology, early access is expected to allow them to conduct advanced security testing, evaluate emerging threats, and build defensive capabilities before such models become more widely available.

Leading AI developers, rather than releasing frontier models simultaneously to all users, are choosing to provide trusted governments, national security agencies, and strategically important industries with early access to assess potential risks and strengthen safeguards.

The stakes have risen considerably over the past year as AI systems have demonstrated growing proficiency in software engineering and cybersecurity tasks. Researchers have warned that advanced models can accelerate vulnerability discovery, automate portions of penetration testing, and dramatically reduce the expertise required to conduct sophisticated cyber operations. This has prompted fears that AI could lower barriers to entry for cybercriminals while increasing the scale and frequency of attacks.

Japanese authorities appear determined not to be caught unprepared.

Katayama said the Japanese government and financial institutions are also expected to gain access to Anthropic’s frontier AI model, Mythos, which has drawn global attention because of concerns over its advanced cybersecurity capabilities. Earlier this month, Japan established a public-private working group dedicated to examining the risks that frontier AI systems could pose to the country’s financial sector.

The initiative signals a growing recognition that cybersecurity is becoming one of the most important battlegrounds in the AI era. Financial regulators are increasingly concerned that traditional security frameworks may struggle to keep pace with the rapid improvement of AI systems capable of generating code, analyzing networks, and identifying vulnerabilities.

The agreement also highlights Japan’s broader ambition to become a leading participant in the global AI ecosystem. While much of the world’s frontier AI development remains concentrated in the United States and China, Tokyo has been working to ensure Japanese institutions gain access to the most advanced technologies while maintaining safeguards against their misuse.

OpenAI’s decision reportedly followed discussions between Japanese and U.S. officials and mirrors similar arrangements the company has pursued with selected organizations in Europe. The move suggests access to frontier AI models is becoming an increasingly important strategic asset, particularly for sectors responsible for national economic stability.

The development comes amid growing competition among leading AI companies, including OpenAI, Anthropic, and Google, to position their models as trusted platforms for governments and enterprises. Security capabilities are emerging as a key differentiator as organizations seek tools that can help defend against increasingly complex cyber threats.

For Japan’s banking sector, the initiative represents more than a technology upgrade. It is part of a broader effort to prepare for a future in which artificial intelligence serves as both a powerful defensive tool and a potentially disruptive force.