TikTok's Chinese parent company ByteDance has figured out a way to access NVIDIA's latest AI chips despite export restrictions, according to a report by The Wall Street Journal. The company is working with a firm called Aolani Cloud and building out Blackwell computing systems in Malaysia.

This should give ByteDance access to around 36,000 B200 chips. That's NVIDIA's most powerful processor. The hardware buildout will reportedly cost more than $2.5 billion. The company says it plans on using this new computing power for AI research and development outside of China.

The country has been unable to access the B200 chip, as it was designed in California and, as such, subject to US export controls. This has led some Chinese companies to do what ByteDance is doing with Aolani Cloud. The Singapore-based firm will buy up the components from NVIDIA and will operate exclusively in Malaysia, giving ByteDance access in the process.

"By design, the export rules allow clouds to be built and operated ​outside controlled ​countries," an NVIDIA spokesperson said. They also said that all of the company's cloud partners go through review before being approved to receive its products.

A representative from Aolani Cloud told Reuters ‌that the company adheres to all applicable export control regulations and that ByteDance will be just one of many customers. It plans on providing cloud-computing services to multiple companies across Asia and the globe. However, it's worth noting that Aolani currently operates ‌with just $100 million worth ⁠of hardware and ByteDance is planning to inject a whopping $2.5 billion.

The US did recently allow ByteDance to buy NVIDIA's H200 chips, but they've been slapped with a 25 percent tariff. Additionally, the US government mandated that the export license would only be approved if NVIDIA accepted a Know-Your-Customer requirement, which is an attempt to ensure that China's military can't access the chips. NVIDIA has yet to agree to these terms.

NVIDIA is reportedly working on its own open-source AI agent platform, according to Wired. The chipmaker has been pitching the product to enterprise software companies. Reporting indicates it's going to be called NemoClaw, suggesting that the entire industry is going to embrace this whole "claw" naming convention moving forward.

Just like OpenClaw, this will be a platform in which users dispatch AI agents to perform a variety of tasks. However, NVIDIA's effort looks to have an enterprise focus for now. To that end, reporting indicates that companies will be able to access this platform even if their products don't run on NVIDIA chips.

NVIDIA is currently preparing for its annual developer conference next week and Wired has suggested that the company has already reached out to entities like Salesforce, Cisco and Google to strike partnerships for its platform. It's not clear if these discussions have led to anything official, as none of these companies have provided statements.

This could be a steep climb for NVIDIA, as usage of these multi-purpose agents in the enterprise space is relatively controversial. Some tech companies have asked employees to refrain from using OpenClaw and related tools on their work computers, as the agents can be unpredictable and cause all manner of mayhem. A Meta employee recently shared a story about an AI agent going rogue and mass deleting emails.

This poses a serious security risk to enterprise customers. It's one thing if the claw is trapped on a personal computer, but another thing if it has access to an entire enterprise network. NVIDIA is reportedly beefing up NemoClaw with additional layers of security for AI agents, which is likely an effort to attract those business customers.

Why is this a big deal? Unlike traditional chatbots that typically require hand-holding from the user every step of the way, claws are designed to run autonomously on computers and perform complex, multi-pronged tasks without too much human supervision. 

This all started with software originally called Clawdbot, which is now called OpenClaw. The creator of OpenClaw, Peter Steinberger, recently joined OpenAI to help "drive the next generation of personal agents."

Uber has expanded its program that helps pair women riders and drivers. The Women Preferences feature is now available nationwide, after being tested in several cities. It has previously been available in many countries around the globe and started in Saudi Arabia back in 2019.

It's pretty easy to use. Women riders will see an option for Women Drivers when requesting a trip, and this also works when making a reservation in advance. Users can also make a preference for a woman driver in the settings app, though this doesn't guarantee anything and depends on the driver pool.

The feature works in much the same way for drivers. Women drivers will be able to request trips with women riders via the settings.

Uber isn't the only rideshare company trying to make half of the world's population a bit safer during trips. Lyft has been expanding its own take on the feature, which it also recently took nationwide.