Recent court documents reveal that OpenAI had contemplated purchasing Cerebras, an AI chip manufacturer preparing to go public.

According to new evidence in Elon Musk’s continuing case against OpenAI, OpenAI was considering acquiring Cerebras in or around 2017—a year after Cerebras was founded and only a few years after OpenAI started operations.

Ilya Sutskever, a former chief scientist and co-founder of OpenAI, proposed purchasing Cerebras through Musk’s electric vehicle business, Tesla, in an email sent to Musk and OpenAI CEO Sam Altman. Musk had some control over OpenAI’s course at the time and was financially invested in the company.

In September 2017, Sutskever wrote, “In the event we decide to buy Cerebras, my strong sense is that it’ll be done through Tesla,” “But why do it this way if we could also do it from within OpenAI? Specifically, the concern is that Tesla has a duty to shareholders to maximize shareholder return, which is not aligned with OpenAI’s mission. So the overall result may not end up being optimal for OpenAI.”

Sutskever lists a number of Cerebras-related agenda items in a previous email sent in July 2017 to Musk and OpenAI co-founder Greg Brockman, who is currently the company’s president: “Negotiate merger terms with Cerebras” and “More due diligence with Cerebras.”

Although it’s unclear from the exhibits why, the merger deal would eventually fail. Additionally, OpenAI would put its chip goals on hold for years.

Based in Sunnyvale, California, Cerebras creates specialized hardware for AI model execution and training. The company asserts that its chips are quicker and more effective for AI workloads than Nvidia’s flagship products.

Cerebras has received $715 million in venture funding and is apparently looking to use the IPO to almost treble its $4 billion valuation. But it has a lot of obstacles to overcome. Eighty-seven percent of Cerebras’ revenue in the first half of 2024 came from a single Abu Dhabi company, G42, whose longstanding ties to China have alarmed U.S. politicians. Having pled guilty to evading financial controls while serving as a vice president at the publicly traded business Riverstone Networks, Andrew Feldman, the CEO of Cerebras, likewise had a troubled past.

The transaction might have been advantageous to both businesses if it had taken place. OpenAI may have had an important advantage in its competition to develop in-house chips, while Cerebras would have avoided the road to a challenging IPO.

Since Nvidia holds a significant portion of the market for processors with AI optimization, OpenAI has long aimed to lessen its need on the company. OpenAI is under pressure to lower the cost of training, fine-tuning, and operating models, even though it is late to the in-house chip game—Google and Amazon Web Services, among others, have long provided chips made for AI workloads. It could be able to achieve the necessary savings by having its own chips.

At one point, OpenAI was thinking about being an acquisition target and intended to build a network of chip manufacturing factories. However, IBM has apparently shelved such plans in favor of rapidly expanding a team of engineers and chip designers and collaborating with semiconductor companies TSMC and Broadcom to develop an AI processor for running models. It might show up as early as 2026.

Starfish Space, a business providing satellite services, revealed on November 13 that it has raised $29 million in a fresh round of funding headed by Shield Capital. Munich Re Ventures, Toyota Ventures, NFX, and Industrious Ventures are among the other investors in the round, along with newcomers Point72 Ventures, Booz Allen Ventures, Aero X Ventures, Trousdale Ventures, and TRAC VC.

Established by former engineers from Blue Origin and NASA, Starfish Space creates self-sufficient satellite maintenance vehicles to prolong the lifespan of satellites and eliminate space junk. Recent successes for the company include agreements with NASA and commercial satellite operator Intelsat, as well as a $37.5 million contract with the U.S. Space Force.

With the closing of the latest round, Starfish has raised more than $50 million in total fundraising to date.

Otter is an in-space maintenance vehicle created by Tukwila, Washington-based Starfish. Starfish will be able to finish developing the first three Otter vehicles with the new money, which will be used for missions for NASA, the U.S. Space Force, and Intelsat. In 2026, Intelsat and the U.S. Space Force are expected to launch their Otter missions into geostationary orbit.

The investment in Starfish is the third space-focused investment made by Booz Allen Ventures, the startup capital division of consulting behemoth Booz Allen Hamilton. According to Chris Bogdan, executive vice president of Booz Allen and head of the company’s space division, “This investment aims to strengthen the resilience and sustainability of space infrastructure through innovative offerings for both government and commercial mission sets,”

Prior space investments made by the corporation include Quindar, which automates satellite fleet management operations, and Albedo, which creates low-flying satellites for high-resolution Earth observation.Starfish Space, a business providing satellite services, raises $29 million.

A metal transition lies at the core of the energy transition. Compared to their gas-powered counterparts, wind farms, solar panels, and electric cars require a lot more copper, zinc, and nickel. Additionally, they need additional rare earth elements—exotic metals with special qualities—which are necessary for the magnets used in devices like EV motors and wind turbines.

China now controls the majority of rare earth element processing, purifying around 60% of the world’s supply. The Biden administration has stated that the scenario presents challenges to national and economic security, as demand for these minerals is expected to soar.

In the United States and many other countries, large amounts of rare earth metals are currently sitting untapped. The problem is that they are combined with a ton of hazardous mining waste.

Phoenix Tailings is expanding a method for extracting elements from mining waste, such as nickel and rare earth metals. After collecting oxidized metal with water and recyclable solvents, the company heats a mixture of molten salt and applies electricity to the metal.

Co-founded by MIT alums, the business claims that its pilot production plant in Woburn, Massachusetts, is the only location in the world that produces rare earth metals without emitting carbon dioxide or hazardous byproducts. Phoenix Tailings now uses renewable energy contracts to offset the electricity used in the process.

By 2026, the company anticipates producing over 3,000 tons of the metals, which would have accounted for almost 7% of all U.S. output in the previous year.

Phoenix Tailings is now increasing the range of metals it can manufacture and moving forward with plans to construct a second manufacturing plant with help from the Department of Energy.

According to the founding team, which consists of Nick Myers, Anthony Balladon, and MIT graduates Tomás Villalón ’14 and Michelle Chao ’14, the work has global and geopolitical ramifications.

“Being able to make your own materials domestically means that you’re not at the behest of a foreign monopoly,” Villalón explains. “We’re focused on creating critical materials for the next generation of technologies. More broadly, we want to get these materials in ways that are sustainable in the long term.”

Addressing a worldwide issue

After enrolling in Course 3.091 (Introduction to Solid-State Chemistry) during his first year at MIT, Villalón developed an interest in chemistry and materials science. He had the opportunity to work at Boston Metal, another MIT startup that decarbonizes steel production on a large scale using an electrochemical technique, during his senior year. Villalón, a materials science and engineering major, began considering developing more environmentally friendly metallurgical techniques as a result of the event.

But Villalón didn’t take action until he happened to meet Myers at a Bible study in 2018.

When the subject of electricity came up, “We were discussing some of the major problems in the world when we came to the topic of electrification,” Villalón remembers. It turned into a debate about how the United States obtains its materials and how we ought to consider electrifying their manufacturing. After ten years of working there, I eventually thought, “Let’s go do something about it.” Nick concurred, but I assumed he was merely trying to boost his self-esteem. Then, in July, he called me at random and said, ‘I’ve got [$7,000]. When do we start?’”

The founders began testing novel methods for making rare earth metals after Villalón brought in Chao, a former MIT classmate and fellow materials science and engineering major, and Myers brought in Balladon, a former coworker.

According to Villalón, “We went back to the base principles, the thermodynamics I learned with MIT professors Antoine Allanore and Donald Sadoway, and understanding the kinetics of reactions,”  “Classes like Course 3.022 (Microstructural Evolution in Materials) and 3.07 (Introduction to Ceramics) were also really useful. I touched on every aspect I studied at MIT.”

The founders also participated in the U.S. National Science Foundation’s I-Corps program and were mentored by MIT’s Venture Mentoring Service (VMS). Sadoway advised the business.

The inventors constructed a prototype reactor in Villalón’s backyard after creating a preliminary version of their system design and purchasing an experimental amount of red sludge, a mining waste. In the end, the founders had a modest amount of product, but they had to quickly borrow the scientific tools necessary to identify it. It turned out to be pure iron and a trace amount of rare earth concentrate.

Today, Phoenix Tailings warms its combination to about 1,300 degrees Fahrenheit at its refinery in Woburn, where it incorporates mining waste that is rich in rare earth metals. Pure metal gathers on an electrode when an electric current is applied to the mixture. There is not much garbage left over after the operation.

Because rare earths require extremely high purities in comparison to metals manufactured traditionally, Villalón says, “the key for all of this isn’t just the chemistry, but how everything is linked together.” “As a result, you have to be thinking about the purity of your material the entire way through.”

Rare earths, nickel, magnesium, and other elements

When using renewable energy sources to generate power, Villalón claims the process is 100% carbon free, creates no harmful byproducts, and is cost-effective when compared to traditional manufacturing methods.

Neodymium and dysprosium, two rare earth elements crucial to magnets, are now produced for clients at the Woburn site. Consumers are utilizing the materials for defense applications, electric vehicles, and wind turbines.

Additionally, the business has been awarded two grants totaling over $2 million under the U.S. Department of Energy’s ARPA-E program. Its 2023 award funds the creation of a technology that employs carbonization and recycled carbon dioxide to extract nickel and magnesium from mining waste. Magnesium and nickel are both essential components for clean energy devices like batteries.

The company will use the most recent funding to modify its method so that it can generate iron from mining waste without emitting any harmful byproducts or emissions. Phoenix Tailings claims that it has an abundance of material to work with and that their technique is suitable with a broad range of ore types and waste materials: About 1.8 billion tons of garbage are produced annually in the United States as a result of the mining and processing of mineral ores.

Villalón says, “We want to take our knowledge from processing the rare earth metals and slowly move it into other segments,”  Here, “We simply have to refine some of these materials here. There’s no way we can’t. So, what does that look like from a regulatory perspective? How do we create approaches that are economical and environmentally compliant not just now, but 30 years from now?”