This big AI bottleneck could be solved using light

The AI ​​boom is a hype cycle like no other in many ways. Sure, you have to compare the dot-com rise of the late 90s to the mobile revolution of the 1990s, but the company leads the pack in terms of capital invested and high expectations for creating major changes in society.

The speed of progress comes with major hurdles. AI builders must contend with constraints such as access to energy to power huge data centers, a shortage of memory chips, and efficient data transfer between AI chips and systems.

An emerging technology known as photonics offers a path to solving the latter.

Photonics can be used in AI infrastructure by using light to move data between graphics processing units (GPUs), memory, networking chips, servers, and data centers, rather than relying on electrical signals flowing along copper wires. Some photonics technologies are already in use, such as fiber optic connections.

However, currently, many of the connections between AI servers and within racks are made along copper wires, which limits speed and increases energy costs.

“One of the main bottlenecks in AI model performance is the communication speed between chips and chip servers,” said Gil Luria, head of technology research at DA Davidson.

“The faster the communication, the faster users can get answers and perform tasks,” he added. “Moving the chip-to-chip and server-to-server connections to optical connections significantly improves the performance of the model.”

The potential of photonics has become visible Nvidia Pour billions of dollars into companies that develop the technology.

Since early March, Nvidia has announced $2 billion in investments. lumen, reasonable and marvelboth of which are developing photonics technology. The semiconductor giant also announced an investment of $500 million. corning The company aims to develop advanced optical connectivity solutions and participated in optical startup Ayar Labs’ $500 million Series E funding round.

Nvidia CEO Jensen Huang told an audience at GTC in March that the semiconductor giant is starting to expand its silicon photonics technology.

“We need significantly more silicon photonics technology capability than the world has today,” he said. Huang added that Nvidia has begun deploying photonics on its networking platform and GPU-to-GPU interconnection platform.

However, implementing new technology at scale is never a smooth process.

Alan Weckel, principal analyst at market research firm 650 Group, told CNBC that one hurdle will be manufacturing. “The industry has never seen demand or growth like this before, so strengthening the supply chain to meet demand is difficult, especially when there are constraints.”

Another is adapting active AI systems to photonics technology, Luria said.

“The main challenge in incorporating more optical components is that existing product roadmaps will need to be significantly redesigned to a different configuration that uses copper instead of fiber,” he said.

“It may take another generation or two of products from Nvidia and others to become more widespread.”

snowflake Recorded best day ever after strengthening guidance and announcing deal with AI Computing Amazon on Thursday.

French startup Mistral AI is considering designing its own chip and may eventually develop one, CEO Arthur Mensch told CNBC.

SK Hynix’s market capitalization surpassed $1 trillion on Wednesday, and its stock price rose as much as 11% as investors continued to flock to artificial intelligence-related semiconductor stocks.

Taiwanese chip stocks rose after Nvidia announced a $150 billion spending plan in the country on Wednesday.

Blue Origin’s New Glenn rocket exploded Thursday night during a high-temperature fire test at the Space Force’s launch facility in Cape Canaveral, Florida. Jeff Bezos said all staff members were safe after the explosion.

US memory chip manufacturer micron The company surpassed the $1 trillion market cap for the first time on Tuesday after its stock price soared 19%. The company anticipates huge demand amid memory shortages and the AI ​​boom. The company’s stock price has increased nearly 200% so far in 2026.