Nvidia and Foxconn are building Taiwan’s largest supercomputer using Nvidia Blackwell chips.

The project, Hon Hai Kaohsiung Super Computing Center, revealed Tuesday at Hon Hai Tech Day, will be built around Nvidia’s Blackwell graphics processing unit (GPU) architecture and feature the GB200 NVL72 platform, which includes a total of 64 racks and 4,608 Tensor Core GPUs.

With an expected performance of over 90 exaflops of AI performance, the machine would easily be considered the fastest in Taiwan.

Foxconn plans to use the supercomputer, once operational, to power breakthroughs in cancer research, large language model development and smart city innovations, positioning Taiwan as a global leader in AI-driven industries.

Foxconn’s “three-platform strategy” focuses on smart manufacturing, smart cities and electric vehicles. The new supercomputer will play a pivotal role in supporting Foxconn’s ongoing efforts in digital twins, robotic automation and smart urban infrastructure, bringing AI-assisted services to urban areas like Kaohsiung.

Construction has started on the new supercomputer housed in Kaohsiung, Taiwan. The first phase is expected to be operational by mid-2025. Full deployment is targeted for 2026.

The project will integrate with Nvidia technologies, such as Nvidia Omniverse and Isaac robotics platforms for AI and digital twins technologies to help transform manufacturing processes.

“Powered by Nvidia’s Blackwell platform, Foxconn’s new AI supercomputer is one of the most powerful in the world, representing a significant leap forward in AI computing and efficiency,” said Foxconn vice president James Wu, in a statement.

The GB200 NVL72 is a state-of-the-art data center platform optimized for AI and accelerated computing.

Each rack features 36 Nvidia Grace CPUs and 72 Nvidia Blackwell GPUs connected via Nvidia’s NVLink technology, delivering 130TB/s of bandwidth.

Nvidia NVLink Switch allows the 72-GPU system to function as a single, unified GPU. This makes it ideal for training large AI models and executing complex inference tasks in real time on trillion-parameter models.

Taiwan-based Foxconn, officially known as Hon Hai Precision Industry Co., is the world’s largest electronics manufacturer, known for producing a wide range of products, from smartphones to servers, for the world’s top technology brands. Foxconn is building digital twins of its factories using Nvidia Omniverse, and Foxconn was also one of the first companies to use Nvidia NIM microservices in
the development of domain-specific large language models, or LLMs, embedded into a variety of internal systems and processes in its AI factories for smart manufacturing, smart electric vehicles and smart cities.

Water companies have been ordered to return £158m to customers via lower bills next year after missing key targets on issues like pollution.

The industry regulator Ofwat announced the rebate following its annual review of water and wastewater companies’ performance in England and Wales.

Chief executive David Black warned companies that “money alone” would not solve the issue and there needed to be a change in culture.

Industry body Water UK have been approached for a comment.

Ofwat assesses the performance of the 17 largest water and wastewater companies in England and Wales each year against key targets for issues such as sewer flooding, supply interruptions and water leaks.

For the second year in a row, no company achieved the top rating, although four companies did show an improvement from last year.

“Companies must implement actions now to improve performance… and not wait till government or regulators ask them to act,” David Black said.

For the worst performing companies, failure to meet the targets means they will now have to collectively return £157.6m to customers on their bills for 2025-2026.

However, this is only expected to amount to a reduction of a few pounds on bills and will be dwarfed by the expected long-term increase of £94 per household over the next five years, which Ofwat is currently proposing.

This price rise is still being negotiated with water companies, who argue that even higher bills are needed if they are to afford the infrastructure improvements required to tackle some of the issues raised in this performance report.

French biotech startup Generare has closed a €5 million seed round (around $5.5 million at current exchange rates) to step up development of what it touts as a highly scalable approach to identifying promising compounds for drug discovery that already exist in nature. It’s focused on sifting for molecules that are produced by microorganisms but could be repurposed as treatments in humans.

Such molecules were once a rich source of drug discovery using early chemistry-based discovery techniques. The cardinal example is Penicillium’s antibiotic effect that was spotted after the mold happened to grow in a lab petri.

But finding less common molecules that could become the basis for new antibiotics, anticancer or other drugs requires a new kind of approach that can sift through large amounts of genetic material, says Generare co-founder and chief science officer Dr Vincent Libis.

“We are really interested in discovering chemical molecules produced by bacteria,” he explains. “They encode them in genes — which is basically [a] genetic recipe for [a] molecule. And so what we are hunting for is these genetic recipes. And our technologies are all focused on detecting novel genetic recipes and then manipulating them to obtain the molecules that they encode.”

“So [it’s] a lot of molecular biology to sequence DNA, to cut and paste DNA, and bioinformatics, or computational biology, to triage which one of these genetic recipes you want to go after first, and what do you expect the molecules that they encode are gonna be like.”

Drug hunting ‘on a planetary scale’

Founded in October 2022, the startup says it’s devised a technique that relies upon cloning and biosynthetics to break up the genetic material of microorganisms into millions of fragments to allow for speedier analysis — making it much quicker and easier to identify new molecules of interest.

Although still young for a biotech, Generare has already been able to identify more than 1,000 “genetic recipes”, as it calls the molecules of interest, since it began operating its discovery platform 12 months ago. Part of this early haul includes more than 100 “novel” chemical molecules with a handful that have antibiotic activities.

The approach it’s taking depends on scaling this up to cast a very wide net so it can crunch through masses of genetic material to find the minority of useful compounds hiding amid the dirt.

“Because these molecules are very successful, the more you find, the more chances you have of finding the next billion dollar drug, basically,” CEO and co-founder Guillaume Vandenesch tells TechCrunch. “It’s like you’re playing a board game and you’re trying to get a double six — but you’re throwing like 100 or 1000 dice at a time.

Generare is drawing on around a decade of work by Libis in genetic engineering which underpins its approach. Essentially the method involves taking DNA extracted from a microorganism that codes for a natural molecule, putting the genetic recipe into a lab host and producing (he calls it printing) the molecule in a test tube — “where you can really interrogate it”.

The startup says this method allows it to explore the entire chemical diversity of microbes. It also talks about its platform enabling scanning for novel genetic recipes “on a planetary scale” as a result of how efficiently it’s able to conduct the search.

“If you want to explore the entire map of natural molecules that are produced by microbes you need to have a technology that is extremely efficient, extremely cost efficient,” stresses Vandenesch.

DNA sequencing has been used in drug development for around 15 years, per Libis. More recently, about five years ago, a wave of startup began using DNA sequencing to mine natural molecules specifically. Generare is aiming to improve on earlier efforts by accelerating the analysis of the genetic material.

It’s patenting its approach and what Libis refers to as “the most mathematically efficient implementation” of it.

“What we bring to the table, the really differentiating technology, is at which scale we can do this,” he tells TechCrunch. “We really have orders of magnitude faster ability to get from a genetic recipe — unknown — to a molecule in a test tube. And this is enabled by our, let’s call it, cloning technology, which just pieces out the strands of DNA that contains this genetic recipe very quickly into a laboratory strain so that we can express and then characterize the activity.”

He likens the process to removing a blindfold which reveals the genetic diversity of the molecules. By doing that at scale the startup can organize and group compounds into groups which looks the same; groups which make already patented molecules (which it avoids); and groups which look like they might encode promising molecules. “So you have a much more orderly search that is enabled by this — by using DNA as a guide,” he adds.

“I worked for pretty much 10 years on finding a solution to the speed at which we could take a piece of DNA and bring it to another laboratory strain, which was the bottleneck of this whole process,” Libis continues. “We stumbled on a solution — where we gained orders of magnitude faster ability to transfer these genes. And so that’s what motivated the start of the company.

“That’s our really differentiating element… So it’s just a lot of research,” he adds.

Digging in the dirt

Soils are Generare’s initial choice for scaling this natural molecules search on account of how many microorganisms they contain and the ancient “war” of survival/supremacy that’s been waged between the countless microscopic entities living in the ground. It encourages the development of novel chemical protections against things like bacteria or fungi or other threats.

“In the soil, you have 1,000 species [of bacteria] per gram and it’s on the whole planet,” says Libis, adding: “This war is [waged] through chemical weapons — which are these molecules — and so they’ve been ‘innovating’, the war has been going on for hundreds of millions of years.”

Identifying more of the “really powerfully chemistry” that’s evolved through natural processes, whereby one biological entity develops a tool “to mess with the biology of another organism”, and putting it to “good use” in the human body is Generare’s core mission.

“These bacteria evolved for 400 million years. They’ve probably invented a lot of things that could be very valuable to us,” adds Vandenesch. “There’s so many different organic organisms that are adapted to so many different conditions… basically there is a treasure trove that’s ready to be discovered.”

Beyond soil microbes, the startup says marine environments offer another vast hunting ground where it can apply its approach.

“Really the exciting thing, in a way, is we know only 3% of all these chemicals that are made by microbes,” Libis adds.
“[Humanity has] only discovered 3%… so a large reservoir still needs to be explored.”

“We know that we only found 3% because we see 97% more genes than we see known molecules.”

Industrializing a decade of research

So far Generare has secured an agreement with French biotech company Aurobac Therapeutics to pilot its technology. It’s a company that’s focused on developing new antibiotics.

The new seed funding will be used to industrialize its gene transfer approach so it can figure out how to scale up by turning its lab-based processes into more of a factory-style production line — that’s as lean and streamlined as possible.

“The vision is to go planetary scale exploration of what used to be the most successful modality in the history of pharma and agrochemistry as well,” says Libis. “But basically, the next two years, we are just trying to make it ready for the prime time — where we go after the whole world, basically.”

Its business model will see it engaging in “co-development” of drugs with pharmaceutical companies — with the goal of earning a return on any “valuable bioactive molecules” it has brought to the attention of its partners.

Given the long timescales involved in bringing new drugs to market, Generare confirms the startup will be looking to raise further funding, likely in a couple of years.

“Our dream is like, in 18 months, we have the blueprints of the faculty, you know, meter per meter, exactly what each step is doing and what equipment is needed,” he adds. “Then that’s what we will be fundraising for — to build this in real life.”

The company also believes its platform could also have applications in other sectors, such as supporting development of new agri-chemicals with natural molecules to replace traditional pesticides and insecticides. There could even be uses in the cosmetics industry. But its initial focus is on pharmaceuticals where there is a pressing need.

Antibiotics resistance is a growing problems so there’s a clear need to rethink drug discovery. However Libis also warns the key issue is the broken business model which needs governments to step in with subsidies if humanity is to avoid a full-blown crisis of treatment failure.

“Antibiotics are just a broken market,” he says. “Economically, it’s very hard to have a viable business model when you discover antibiotics. It’s such a commodity once you have them. And also, people don’t use the most powerful ones, so companies that discover them have a hard time getting back their R&D expenses. So what we need is, in fact, mostly a governmental answer — at the EU scale, at the US scale — that will inject some guarantee that the company that invests in R&D and antibiotics will get a return on their investment.”

“There’s hope,” he adds. “There’s the PASTEUR Act in the U.S. that might pass and offer this kind of guarantee, and the EU as well is working actively. But it’s really urgent that something happens there. So for now, we are taking the risk, and we managed to convince our investors that we should be working on this activity, that the landscape is going to change and that there’s going to be, like, subsidies to compensate. But it’s daily struggle.”

“For two years we will manage to bend their mind. But in five years, if nothing has changed at some point we will be forced to also refocus on more guaranteed indications.”

Investors in Generare’s seed include Teampact.ventures, Galion.exe and EU-backed VIVES Partners. Synbioven, Saras Capital and Better Angle also participated, along with a number of business angels. 

With so many molecules lurking somewhere out there waiting to be discovered, Generare doesn’t sound worried about competitors raiding nature’s treasures before its had a chance to apply its high scale sifting methodology. But it name-checks the likes of U.S.-based Hexagon Biosciences and LifeMine Therapeutics as playing in the same arena.

Hurricane Milton strengthened into a Category 5 storm today, becoming one of the most rapidly intensifying storms ever in the history of the Atlantic hurricane season. 

“This is almost like three times the threshold that is used. So, yeah, this is definitely off the charts,” Karthik Balaguru, a climate scientist at the Pacific Northwest National Laboratory, says. Only Hurricane Wilma in 2005 and Hurricane Felix in 2007 have strengthened more rapidly than that, the NHC says.

Rapid intensification is becoming a bigger risk with climate change. Storms that gain strength that quickly can give communities less time to prepare. Milton is headed toward the west coast of Florida, where many residents are still recovering from Hurricane Helene’s wrath.

Milton’s sustained winds were estimated to be as high as 160 miles per hour by 11:55AM ET on Monday, October 7th, according to data from an Air Force Reserve Hurricane Hunters aircraft. That puts it in the strongest category of storms — Category 5 — according to the Saffir-Simpson hurricane scale. Wind speeds had reached a whopping 175mph by the NHC’s next update at 2PM ET. Milton has also managed to strengthen from a Category 1 to Category 5 storm at the second-fastest rate on record for the Atlantic, tying Hurricane Maria.

Hurricanes draw strength from heat energy at the surface of the sea. Unusually warm waters in the Gulf of Mexico helped supercharge Milton, as they did for Helene less than two weeks ago. Both storms rapidly intensified as they approached shore, benefitting from low wind shear that might otherwise tear a storm apart before it strengthens.

“This has been happening. Whether it is the third most or the 10th most [rapidly intensifying storm], it shouldn’t matter,” Balaguru says. “[Milton] fits this pattern of storms intensifying more rapidly with climate change. I think that is something for people to think about, especially when it happens like this, close to landfall.”

The storm is forecast to move near Mexico’s Yucatan Peninsula today before approaching Florida on Wednesday. The storm surge could cause up to six feet of flooding along parts of the northern coast of the Yucatan Peninsula. Florida’s Tampa Bay could potentially see a catastrophic 12-foot storm surge. Milton could meet stronger wind shear before hitting Florida’s Gulf Coast, weakening the storm. But it’s still projected to make landfall as a major hurricane “with life-threatening hazards.”

Hurricane Helene made landfall in Florida’s Big Bend region on September 26th as a Category 4 storm, bringing a 15-foot storm surge to the area before tearing a devastating path across southeastern states up to North Carolina and leveling entire communities along the way. Milton could prove to be even more dangerous to Florida, in particular, as it barrels toward more populated regions along the state’s western coast.