H&M wants to make clothing from CO2 using this startup’s tech; plus, Niv-AI exits stealth to wring more power performance out of GPUs

0:00This is TechCrunch. This episode is brought to you by Framer. Set the tone for your brand by creating the dot com of your dreams with Framer, a website builder that makes sure you're not leaving opportunities on the table. Framer can turn your website into a tool for growth by helping you build a better one faster. Framer is like your favorite design tool. With real-time collaboration, a robust CMS with everything you need for great SEO and advanced analytics that include integrated AB testing, your designers and marketers will feel confident that they're helping your business move the needle. It's the shortcut you've been looking for. And then when your website looks exactly as you wanted to, changes to your Framer site will go live to the web in seconds when you click publish. It without help from engineering. Here's how I think Framer could be helpful to you. If you normally have to wait on someone else to make changes to your website, you can

1:02cut that out of your workflow and get these changes live quickly. Upgrade your website and give your business a boost. Learn how you can get more out of your dot com from a Framer specialist or get started building for free today. At Framer.com slash TC Startups for 30% off a Framer Pro angle plan. That's Framer.com slash TC Startups for 30% off. Framer.com slash TC Startups rules and restrictions may apply. You don't have to tell the fashion industry that it has a waste problem. It already knows about one garbage truck of textiles is thrown away every second. Meanwhile, the industry generates more carbon pollution than international flights and maritime shipping combined. Some companies are experimenting with new ways to recycle textile waste, while others are developing new materials that will not require fossil fuels. One such company, a startup named Ruby, is basically taking the machinery of biology outside

2:05of the cell to make the building blocks of Lyosel and viscos. That's according to the company's co-founder and CEO, Nika Machufe. The startup's technology would allow any company that uses cellulose to build products from captured carbon dioxide. Ruby recently raised $7.5 million to build a demonstration scale of its cellulosic production system, which is designed to produce tens of tons of material using CO2 as its main ingredient. The round was led by AP Ventures and FH1 investments with participation from CMPC Ventures H&M Group, TALUS Capital, and Understory Ventures. The startup has booked more than $60 million in non-binding off-take agreements with a couple of partners. The company has tested the material with 15 pilot partners, including H&M, Patagonia, and Walmart. To make cellulose from Lyosel or viscos, Ruby uses enzymes, that differs from other startups,

3:10which might use engineered bacteria inside a fermenter, or chemical catalysts to transform carbon dioxide into the compound. Most cellulose comes from trees, including plantations and virgin rainforests. Machufe said these textile and raw material supply chains are very long. Here in the U.S., we've gotten interest in being able to actually produce cellulose pulp that's textile grade, where that doesn't exist today. The idea to use enzymes came when Machufe, who as a scientist researched new materials, teamed up with her twin sister, Lila, who was studying medicine at Harvard Medical School. She said, we looked at all the tech out there, but they kept coming back to enzymes. The enzyme industry is massive, she said. It's used to make high fructose corn syrup and in wastewater treatment. Ruby uses a cascade of enzymes to process waste carbon dioxide.

4:11The company has used AI and machine learning methods to boost the enzymes efficacy and stability. Currently, the enzymes float in an aqueous solution, and as carbon dioxide is added, white cellulose will appear inside the reactor within a few minutes. The reactors fit inside shipping container-sized modules. Eventually, Ruby plans to change its process to allow for continuous production. While the startup is targeting apparel companies, as its first customers, eventually it hopes to provide cellulose to any industry that uses it. Machufe said, this really is a platform. We think of it as a platform to make all the important chemicals and materials across the economy in a low-cost way. Technology is a key raw material for artificial intelligence, but new processing techniques outstrip the ability of data center operators to manage their relationship with the power

5:12grid, forcing them to throttle down by as much as 30%. Nvidia's CEO, Jensen Wang, said during a recent keynote speech that there is so much power squandered in these AI factories. Every unused watt is revenue lost. Well, startup Niv AI has emerged from stealth this week with $12 million in seed funding to solve this problem by precisely measuring GPU power use with new sensors and developing tools to manage it more efficiently. The Tel Aviv-based startup was founded last year by CEO Tomor T. Moore and CTO Edward Cazise and is backed by Grove Ventures ArcVC and Coded VC leap forward among others. The company declined to share its valuation though. As Frontier Labs operate thousands of GPUs in concert to train and run advanced models, there are frequent millisecond scale power demand searches as the processors switch between

6:17computation tasks and communicating with other GPUs. These searches make it difficult for data centers to manage the power they draw from the grid. To avoid being left without sufficient electricity, data centers pay for temporary energy storage to cover searches or throttle their GPU usage. Both cases reduce the return on investments and expensive chips. Theore Handelsman, a partner at Grove Ventures, who sits on Nibb's board, said, we just can't continue building data centers the way we build them now. The first step in Nibb's roadmap is understanding what's going on. The company is now deploying rack level sensors that detect power usage at the millisecond level on GPUs that it owns and alongside design partners. Now the goal is to understand the specific power profiles of different deep learning tasks and develop mitigation techniques that allow data centers to unlock more of their existing capacity.

7:21Naturally, the engineers expect to build an AI model on the data they collect, with the goal of training it to predict and synchronize power loads across the data center, a co-pilot for data center engineers. Nibb's AI expects to have an operational system in a handful of US data centers in the next, say, six to eight months, it's an attractive idea as hyperscalers trying to build new data centers face difficult land use and supply chain hiccups. The founders see their ultimate product as a missing intelligence layer between data centers and the electrical grid. Timor told TechCrunch, you know, the grid is actually afraid of the data center consuming too much power at a specific time. The problem we're looking at is a problem with two sides of the rope. One is to try to help the data centers utilize more GPUs and hopefully make more of the power that they're already paying for. On the other hand, you can also create much more responsible power profiles in between

8:23the data centers and the grid. That's all for now. For more stories like these, go to TechCrunch.com. This podcast is brought to you by Carvana. Car shopping shouldn't feel like preparing for a marathon of paperwork. That's why Carvana makes buying and financing your car easy. From start to finish, search thousands of vehicles with great prices, all online, all on your time. And when you're ready, your new car shows up right at your door. It doesn't get better than that. Buy your car the easy way. Carvana. Delivery fees may apply.