WuBlockchain’s WhiteLine Daily on Friday laid out four AI industry updates covering financing, cloud capacity, storage earnings, and model safety.
Higgsfield adds NVIDIA and TCV in extended Series B round
U.S. AI video and image generation platform Higgsfield said on Oct. 23 that it had extended its Series B financing, with NVIDIA and technology investment firm TCV joining as new investors. The company did not disclose the size of the additional financing or its latest valuation.
Higgsfield said the platform has passed 40 million users across 238 countries and regions. Its annualized revenue run rate exceeded $1 billion in September, roughly 18 months after launch. The company also said enterprise revenue grew more than 100% quarter over quarter, enterprise customer contract revenue has increased 10x since June, and the platform now processes about 12 trillion tokens and completes about 150 million content generations each month.
On infrastructure, Higgsfield said it is expanding compute infrastructure to NVIDIA Vera Rubin and using NVIDIA Dynamo to optimize inference deployment. The report noted that annualized revenue run rate is based on recent revenue and does not represent revenue actually realized over the past year.
WhiteLine Daily said the 10x increase in enterprise contract revenue since June and roughly 150 million monthly generations suggest commercial demand has reached a meaningful scale, while rising video generation volume will also push inference compute demand higher.
AWS says 2026 large-customer capacity is largely sold out
Amazon CEO Andy Jassy said in an Andreessen Horowitz podcast interview that AWS cloud capacity for large customers in 2026 is largely sold out, most of its 2027 capacity has already been reserved, and a considerable portion of 2028 capacity already has customer commitments.
Jassy said demand is coming mainly from AI labs and enterprise customers. Amazon is still keeping some capacity available for on-demand services such as EC2 and Bedrock. The company had previously planned about $220 billion in capital expenditures for 2026, and Jassy said that even with higher spending, the company could still earn a reasonable return.
He also said AWS can adjust purchases of servers and networking equipment if customer demand changes. Amazon is expected to report third-quarter earnings on Oct. 29.
WhiteLine Daily said long-term reservations reduce the risk of idle newly added capacity, but customer commitments are not the same as recognized revenue. Contract payment terms and the timing of actual deployment will determine how quickly that large capital outlay can be recovered.
Western Digital posts 55% revenue growth in fiscal Q1 2027
U.S. hard drive storage maker Western Digital reported in its fiscal 2027 first-quarter results that revenue for the quarter ended Oct. 2 reached $4.381 billion, up 55% year over year and 17% from the prior quarter.
GAAP gross margin rose from 43.5% a year earlier to 57.8%, an increase of 14.3 percentage points. Operating income came in at $2.046 billion, up 158% year over year, while operating margin improved from 28.1% to 46.7%.
The company said higher storage capacity shipments and better product pricing drove the results. Quarterly free cash flow reached $1.38 billion. For the next quarter, Western Digital forecast revenue of $5 billion to $5.2 billion and said non-GAAP gross margin is expected to rise again to 60% to 61%.
WhiteLine Daily added that Western Digital’s average selling price per TB rose 28% year over year, indicating that the earnings improvement did not come from shipment growth alone. Whether pricing for high-capacity hard drives can keep rising will directly affect future gross margin.
Moonshot AI releases Kimi K3.1 Pro with limits on cyberattack capability
China-based AI company Moonshot AI released Kimi K3.1 Pro on Oct. 24, describing it as the company’s strongest open-weight model to date with 3 trillion-scale parameters. The model supports a 1 million-token context window and comes with a Highspeed version that runs inference at about 2x to 3x the speed of the standard version.
The company said it proactively removed some cybersecurity-related data and reinforcement learning tasks during training to reduce the risk of the model being used in vulnerability attacks. In internal testing, the model’s score on code vulnerability discovery and exploitation tasks fell from 90.5% in a version without those restrictions to 81.0%, but that was still above the previous-generation K3 score of 33.3%.
The new model is already available through Kimi products and API services, and full model weights are scheduled for release before Nov. 4. The report noted that the security test results came from the company’s internal evaluation and have not been independently verified.
WhiteLine Daily said K3.1 Pro still scored 81.0% in internal vulnerability exploitation tests after the company reduced some offensive capability, far above K3’s 33.3%. It said gains in general coding ability also bring more complicated safety risks.
WhiteLine Daily’s main takeaway
The report said Higgsfield is beginning to turn enterprise customers into revenue, AWS has locked in future compute demand ahead of time, and Western Digital has already improved profitability through storage pricing. Those three developments reflect different stages of business realization: customer payments, capacity reservations, and realized profit.
For Kimi K3.1 Pro, the report raised a separate question: as model capability keeps improving, can developers still limit high-risk uses while keeping model weights open.


