Yazılar

TSMC and Chip Design Firms Use AI to Cut Energy Use in Next-Gen Chips

/in /tarafından

The chips powering artificial intelligence consume enormous amounts of electricity, but Taiwan Semiconductor Manufacturing Co (TSMC), the world’s largest contract chipmaker, unveiled new efforts on Wednesday to make them more efficient—by using AI-powered software in the chip design process.

Speaking at a Silicon Valley conference, TSMC showcased strategies it says could boost the energy efficiency of AI chips by as much as 10 times.

Nvidia’s flagship AI servers, for instance, can draw up to 1,200 watts under heavy workloads—comparable to the electricity used by 1,000 U.S. homes if run continuously. TSMC’s approach centers on a new generation of chiplet-based designs, where multiple smaller chips made with different technologies are packaged together to function as a single processor.

To enable these designs, chipmakers are increasingly turning to AI-driven software tools. Partners like Cadence Design Systems and Synopsys debuted new products on Wednesday, built in close collaboration with TSMC. These tools have shown they can outperform human engineers in solving complex design problems—and in a fraction of the time.

“That helps to max out TSMC technology’s capability, and we find this is very useful,” said Jim Chang, deputy director of TSMC’s 3DIC Methodology Group. “This thing runs five minutes while our designer needs to work for two days.”

Still, physical constraints remain. As chips scale up, moving data on and off them via traditional electrical connections is reaching its limits. New approaches, such as optical interconnects to transfer information between chips, must be made reliable enough for deployment in massive data centers.

“Really, this is not an engineering problem,” said Kaushik Veeraraghavan, an engineer at Meta’s infrastructure group during his keynote. “It’s a fundamental physical problem.”

Intel Seeks Investment from Apple Amid Turnaround Push, Report Says

/in /tarafından

Intel has approached Apple about a potential investment, according to a Bloomberg News report on Wednesday citing people familiar with the matter. The discussions, which also cover ways to work more closely together, are still in the early stages and may not result in a deal. Intel shares closed up 6% following the news.

Intel declined to comment, while Apple has not responded to requests.

The report follows Nvidia’s $5 billion investment for roughly a 4% stake in Intel, announced earlier this week. That deal included plans for the companies to co-develop PC and data center chips, though Nvidia will not use Intel’s foundry business to manufacture its chips.

Intel CEO Lip-Bu Tan has been actively pursuing partnerships as part of efforts to revive the struggling chipmaker. Once a dominant force in the semiconductor industry, Intel has fallen behind in the AI boom to rivals like Nvidia and AMD.

Recent backing has boosted investor confidence: Intel secured a $10 billion government stake engineered by the White House, a $2 billion equity investment from SoftBank last month, and Nvidia’s multi-billion-dollar deal. Since mid-August, Intel’s stock has risen more than 40%.

A potential investment from Apple would add another vote of confidence. Apple was once a major Intel customer before shifting to its own custom-designed silicon in 2020. For Apple, a deal could help diversify its reliance on Taiwan’s TSMC, its primary chip manufacturer, particularly given geopolitical tensions with China.

The move would also align with Apple’s broader U.S. strategy, with the company committing $600 billion in domestic initiatives over the next four years. Strengthening ties with Intel could bolster Apple’s relationship with the Trump administration, which has supported efforts to cement U.S. leadership in advanced technology.

The White House said it has not been directly involved in discussions between Intel and Apple, though a spokesperson emphasized: “The taxpayer has an equity stake in Intel succeeding, and the Administration supports iconic American companies like Intel doing what’s best to cement American tech dominance.”

Intel has reportedly reached out to other potential partners as well, as it works to secure outside investment and long-term clients for its manufacturing facilities.

OpenAI Expands Stargate Scope, Eyes Debt Financing to Secure Chips

/in /tarafından

OpenAI is broadening the scope of its massive Stargate infrastructure project, originally unveiled at the White House earlier this year as a $500 billion initiative with partners including SoftBank and Oracle. Executives now say Stargate encompasses nearly all of OpenAI’s work involving data centers and AI chips, stretching beyond the original plan.

Initially conceived as a new entity for mega-scale AI infrastructure, Stargate has since expanded to cover projects predating its January announcement. OpenAI argues that only massive computing systems like Stargate can power the next phase of the AI revolution.

To finance its chip needs, the company plans to adopt creative strategies including debt financing and chip leasing, estimating savings of 10–15% by renting instead of buying GPUs outright. A newly announced partnership with Nvidia—worth up to $100 billion—will provide $10 billion in upfront cash and long-term backing for data center expansion.

CEO Sam Altman, who has long argued that data centers are the lifeblood of AI, said his goal is to reach the point of building “a gigawatt of new AI infrastructure every week.” Speaking at a briefing in Abilene, Texas—home to Stargate’s flagship site—he acknowledged investor concerns about a potential bubble but insisted long-term growth justifies the scale.

The Abilene facility, under construction by Oracle and Crusoe, spans more than 1,100 acres and employs thousands. The site is said to contain fiber optic cable long enough to stretch from Earth to the Moon and back.

Stargate’s rollout has faced delays due to partner negotiations and site selection challenges, according to SoftBank executives. Still, OpenAI, Oracle, and SoftBank this week announced five new U.S. data centers, bringing Stargate’s active projects to nearly 7 gigawatts of the 10 gigawatts originally targeted.

Executives said Microsoft, OpenAI’s longtime sponsor, will not be included in certain Stargate projects, following negotiations to allow OpenAI to partner more broadly.

The company stressed the urgency: demand for ChatGPT and related tools has already forced OpenAI to delay international product launches due to insufficient compute.

Industry experts note that financing remains a major hurdle. Of the roughly $50 billion cost for a new hyperscale data center, about $15 billion covers land and buildings—while the rest goes toward GPUs, which are both costly and in short supply. Following Meta’s example, which secured $29 billion from outside financiers for a Louisiana data center, OpenAI is expected to rely heavily on debt markets to fund its future sites, with Nvidia’s equity stake boosting lender confidence.

Despite bottlenecks in GPU supply chains, Altman maintains that rapid infrastructure buildouts are essential: “We cannot fall behind in the need to put the infrastructure together to make this revolution happen.”