Yazılar

Tsavorite Secures $100 Million in Pre-Orders for Next-Gen AI Chips

AI startup Tsavorite Scalable Intelligence announced it has secured over $100 million in pre-orders from enterprises and cloud providers across the U.S., Asia, and Europe for its next-generation AI chips, designed to scale complex artificial intelligence workloads efficiently.

The company said demand has been particularly strong for its Omni Processing Unit (OPU) — a new compute architecture that integrates CPU, GPU, memory, and connectivity into a single device. This unified design allows the hardware to be reconfigured for diverse applications, addressing challenges in power efficiency, scalability, and cost that have become central to AI infrastructure development.

Founded in 2023 by former Intel and semiconductor industry veterans, Tsavorite aims to deliver its first AI chips and enterprise-class AI systems by next year. These devices will support agentic AI workflows — autonomous, multi-step AI processes that require high efficiency and interoperability between compute layers.

“We’ve built the first truly composable, developer-friendly AI platform that delivers step-change gains in efficiency, cost, and scale from edge to hyperscale,” said CEO Shalesh Thusoo.

The company is partnering with Samsung Foundry, using its SF4X platform solution to fabricate the OPU. Tsavorite declined to reveal its valuation or total funding raised so far, but analysts say the pre-orders highlight surging demand for specialized AI chips that can handle increasingly complex data center workloads.

OpenAI’s Sam Altman Urges U.S. to Expand Chips Act Tax Credit for AI Development

OpenAI CEO Sam Altman on Friday called for the United States to broaden eligibility under the Chips Act’s Advanced Manufacturing Investment Credit (AMIC), arguing that expanding the incentive to include AI data centers, server production, and grid infrastructure is essential for maintaining U.S. leadership in artificial intelligence.

Altman’s comments follow a letter sent by OpenAI’s Chief Global Affairs Officer Chris Lehane on October 27 to White House Office of Science and Technology Policy Director Michael Kratsios, formally requesting that the AMIC cover AI infrastructure beyond semiconductor fabrication.

“The U.S. needs re-industrialization across the entire stack — fabs, turbines, transformers, steel, and much more,” Altman said on X (formerly Twitter). “That will help everyone in our industry, and other industries, including us.”

Altman emphasized that the request was “very different from loan guarantees to OpenAI,” clarifying that the company is not seeking direct federal funding for its operations. Earlier this week, he confirmed that OpenAI had discussed potential federal loan guarantees for chip factory construction, but not for data centers.

OpenAI has pledged to invest $1.4 trillion over the next eight years to expand its computational infrastructure, reflecting the skyrocketing demand for AI models and chips that power applications like ChatGPT.

As AI becomes a cornerstone of global technology competition, the Biden administration faces growing pressure to balance industrial policy and fiscal discipline. White House AI and crypto czar David Sacks recently reiterated that there will be no federal bailout for AI companies, underscoring Washington’s cautious stance despite mounting private-sector investment.

Underwater Cables: The Hidden Arteries of the AI Boom and Global Internet

Deep beneath the oceans lies one of the most crucial — yet least visible — components of modern life: underwater communication cables. Nearly 95% of the world’s international data and voice traffic flows through this vast network of almost one million miles of fiber-optic lines connecting continents.

These cables carry everything from financial transactions and government communications to video calls, cloud services, and AI data transfers. As artificial intelligence grows more data-hungry, investment in subsea infrastructure is accelerating at record speed.

Between 2025 and 2027, global spending on subsea cables is expected to reach $13 billion, nearly double the investment made over the previous three years, according to TeleGeography.

“AI is increasing the need that we have for subsea infrastructure,” said Alex Aime, vice president of network investments at Meta. “Without that connectivity, you just have expensive warehouses.”

Tech giants are now the biggest investors. Meta’s Project Waterworth, a 50,000-kilometer cable linking five continents, will be the longest in the world. Amazon’s Fastnet, connecting the U.S. and Ireland, will deliver speeds equivalent to streaming 12.5 million HD movies simultaneously. Google has funded over 30 subsea systems, while Microsoft has invested in others to bolster its Azure cloud network.

But as global reliance on these cables deepens, so do concerns about security and resilience. Damaged or sabotaged cables can cut off entire nations — as seen when Tonga lost internet access after a volcanic eruption in 2022.

While most damage stems from accidents — fishing nets or dropped anchors — analysts have noted a rise in suspected sabotage near Taiwan and in the Baltic Sea, often coinciding with geopolitical tensions. In response, NATO launched “Baltic Sentry” in early 2025 to protect critical subsea infrastructure.

The U.S. Federal Communications Commission (FCC) has also tightened rules on foreign ownership of cable systems, citing threats from China and Russia. “We’re making it difficult to connect undersea cables directly from the U.S. to adversary nations,” said FCC Chair Brendan Carr.

From the 1850 telegraph line between Dover and Calais to AI-era fiber networks, subsea cables remain the unseen lifeline of global communication — and the quiet battleground of the world’s next digital conflict.