A new era Silicon Valley is about networking—not the kind you find on LinkedIn.
As the tech industry funnels into AI data centers, chip makers are both increasing innovation around technology that connects chips to other chips, and server racks to other server racks.
Networking technology has been around since the dawn of computing, effectively connecting mainframes to share information. In the world of semiconductors, networks play a role at almost every level of the stack—from the connections between the transistors themselves, to the external connections made between boxes or racks of chips.
Chip Giants like Nvidia, Broad, and Marvell already have a well-established network of Bona Fides. But in the AI Boom, some companies are looking for new communication methods that help them speed up the huge amounts of digital data flowing through data centers. This is where lightmatter technology, celestial ai, and Psiquantum, which uses virtual technology to accelerate high-speed computing, come in.
Optical technology, or photography, to have an upcoming moment. The technology was considered “lame, expensive, and equally useful,” for 25 years until the remarkable AI Boom and Psiquantum Cofounder and COPITE COCEENSIC COCEENTER AND ANALYST ISAMEDBOLT. (Shadbolt appeared on the panel last week fighting flames
Some capitalist capinters and institutional investors, hope to catch the next wave of new chips or at least find the right target, including the billions resulting from studies like these that find new ways to speed up the way data is implanted. They believe that traditional interconnect technology, which relies on electrons, simply cannot keep up with the growing demand for High-bandwidth AI WorkAll.
“When you look back at the sacrifice, the networks were really boring to cover, because the packets of bits were changed,” said Ben Bajarin, a long-time tech analyst who serves as the CEO of strong research strategies. “Now, thanks to AI, you have to move heavy workloads, and that’s why you see fast nentos.”
Great power of the chip
Bajarin and others give credit to Nvidia for being fundamental about the importance of communication when it has done more in technology over the years. In 2020, Nvidia spent about $7 billion to acquire the Israeli firm Mellanox Technologies, which makes fast network solutions for servers and data centers. Not long after that, Nvidia bought CUMELUS networks, to offer its Linux-based software system for computer networking. This was a turning point for Nvidia, which clearly indicated that its associated capabilities would be more powerful when combined with other GPUs and included in data centers.
While Nvidia dominates directly integrated GPU stacks, Broadcom has become a key player in on-chip accelerators and network accelerators. The $1.7 Trillion company is working in partnership with Google, Meta, and more recently, Opelai, on chips for data centers. It’s on the cutting edge of Silicon Photonics. And last month, Reuters reported that the study of a new network chip called Thor Ultra, is designed to provide an important system within the data center. “
In an earnings call last week, the semiconductor design arm announced plans to acquire carrier Dreambig for $265 million. DreamBig makes AI Chiplets – small circuits, modules designed to be integrated into large chip systems – in partnership with Samsung. The beginning has an interesting intellectual property … which [is] The most key to scale-up and the communication of the communication areas “said CEE CEENE RENE HAAS in Realivengs Call.
It’s bright
Lightmatter CEO Nick Harris pointed out that the amount of AI billing power he needs now doubles every three months – much faster than the law of more. Computer chips are getting colder. “Whenever you’re in the shape of the biggest pieces of art you can build, all the functionality behind that comes from connecting the chips together,” Harris said.
His company’s approach is cutting edge and doesn’t rely on social media technology. LightMatter builds the silicon photonics that connect the chips together. It calls for making a fast photonic engine for AI chips, essentially a silicon stack connected to low-level communication technology. The startup has raised more than $500 million in the past two years from former investors GV and T. Rowe Price. Last year, its expansion reached $ 4.4 billion.
