Capital · United States

Xiilab to Raise 10 Billion Won Through Convertible Bonds

Credited to TheElec · thelec.net · 6h ago

Artificial intelligence (AI) infrastructure company Xiilab plans to issue 10 billion won in convertible bonds (CBs) to investors including the National Growth Fund. The proceeds will be used to expand its AI data center (AIDC) business, including preparations for growing demand for next-generation graphics processing units (GPUs) such as Vera Rubin.

Xiilab said on Nov. 6 that its board approved the issuance of 10 billion won in its first series of unregistered, unsecured, privately placed convertible bonds. The conversion price is set at 8,106 won per share, with a five-year maturity. The subscription agreement is scheduled for Nov. 10, with payment due Nov. 16.

The bonds carry a 0% annual coupon rate and a 0% yield to maturity, allowing the company to raise funds without interest expenses through maturity. They contain no downward conversion price adjustment, or refixing, provision. Xiilab said it also secured a call option covering up to 50% of the subscription amount to help mitigate dilution for existing shareholders if its share price rises. Investors may exercise their put options starting 30 months after issuance.

The convertible bonds will be subscribed by National Growth Fund sub-funds managed by Korea Investment Value Asset Management, DB Asset Management, Suseong Asset Management and Orion Capital Management, totaling 8 billion won. The remaining 2 billion won will come from Orion Capital Management’s KOSDAQ Venture Fund and other investors. The National Growth Fund is a government-backed investment fund established to support advanced strategic industries, including AI, semiconductors and biotechnology. Xiilab qualifies as an investment target because AI infrastructure is one of its core businesses.

Xiilab plans to use the proceeds to expand its AIDC business from infrastructure deployment into operations. The company aims to broaden its GPU management capabilities through its proprietary AstraGo platform and generate recurring revenue from operations and management services.

CapitalMiddle East & Africa

Qatar sovereign wealth fund and Brookfield back supersized UK data center

Credited to Semafor · semafor.com6h ago

Useful

Qatar’s sovereign wealth fund and Brookfield are backing a proposed 1-gigawatt data center in northern England in what could become the UK’s largest. The buildout at the Drax Power Station is the first investment from Qatar Investment Authority’s Qai and the alternative asset manager’s $20 billion AI infrastructure investment venture. The partners intend to invest $3.4 billion over the next five years.

The project has a head start for already having a power connection, but still requires a green light to begin construction, an anchor tenant, and government approvals, Global SWF noted. The consultancy added that Gulf sovereign investors are adopting distinct strategies: Abu Dhabi’s MGX is deploying external institutional capital, Saudi Arabia’s HUMAIN is combining AI operations with investments, and Qai is partnering with infrastructure specialist Brookfield to develop computing assets.

Original · Semafor

Front

Capital · Europe

Baidu releases ERNIE 5.5 Flash base model weights under a restricted license

Credited to Rackline · Priya Nair · ernie.baidu.com · 8h ago

Baidu on Nov. 5 released the weights for ERNIE-5.5-Flash-Base, a 280-billion-parameter mixture-of-experts language model that uses 14 billion parameters for each token. Its license bars use in the European Union, the United Kingdom and South Korea, and it requires Baidu's written approval once a user's related revenue passes $50 million a year.

The weights are on Hugging Face in FP8 and BF16 formats and are mirrored on ModelScope. The model is the raw checkpoint from the end of pre-training, before any instruction tuning. Baidu pitches it as a starting point for companies and researchers who want to adapt it to their own domains. The company says a post-trained version of ERNIE 5.5 Flash will follow.

The license, version 1.0 and dated Nov. 5, grants rights to run, modify, fine-tune and redistribute the model only inside its defined territory, which is everywhere except the EU, the UK and South Korea. Using the model or its outputs outside that territory is not covered by the grant.

Research, education, evaluation and personal use are free and need no notice. Commercial use is also free up to $50 million of annual revenue from products that rely on the model, measured over the trailing 12 months and added up across entities under common control. Commercial users in that band have to send Baidu a one-time notice before they start and display the words Powered by ERNIE in their product or documentation. They do not have to wait for Baidu to reply.

Above $50 million, users need Baidu's written authorization before any commercial use, and any fees are set in a separate agreement. A user that crosses the line without approval has to stop until approval takes effect. Applying or negotiating does not count as permission.

In practice, a developer or a smaller company can build on the model for free. A large company would need a deal with Baidu, and nobody in the three excluded markets can use it under this license.

Baidu's ERNIE 5.5 Flash Base opens weights with territory and revenue gates

Model scale, architecture and commercial license terms from the Nov. 5, 2026 release

Table: 280 billion total and 14 billion active parameters; 42 expert layers pick 10 of 512; EU, UK and South Korea excluded; above $50 million needs written approval.
Item	Detail
Total parameters	280 billion
Active per token	14 billion
Expert layers	42 layers; each picks 10 of 512 experts
Attention cache	One cache entry per 128 tokens
Training tokens	33.5 trillion
Training context	Up to 128,000 tokens
Serving context	Supports 1 million tokens
Territory	Worldwide except EU, UK and South Korea
Commercial to $50m	Free with one-time notice; show Powered by ERNIE
Commercial above $50m	Needs Baidu written approval first

Source: Baidu ERNIE release and model card Nov. 5, 2026; Hugging Face license v1.0; Rackline cutServing context uses Baidu release wording: supports a 1 million-token context for fine-tuning and inference. The $50m gate is annual related-product revenue over trailing 12 months, summed under common control (license v1.0). Not open source. No Rackline estimate.

The design aims to cut the cost of computing per token and of long prompts. Each of the model's 42 expert layers picks 10 of 512 experts for every token. Most of its attention layers keep one cache entry per 128 tokens rather than one per token, which shrinks the memory needed for long inputs. Baidu trained it from scratch on 33.5 trillion tokens with context up to 128,000 tokens. It says the model can be fine-tuned and served with context of about 1 million tokens.

Baidu supports serving through the open-source SGLang and vLLM engines, including an official SGLang container and its own vLLM plugin. The license does not commit Baidu to provide hosted inference, compute, support or future versions. Those require a separate agreement.

The company published benchmark results comparing the model with three base models of similar size from other labs, all run under the same test setup. It reported strong results on exams, knowledge, math and science, reasoning, multilingual, code and long-context tests. Baidu ran the tests itself.