Invest ideasNews
22 Jan 2026, 15:50
Optical processors breakthrough: Neurophos secures $110M to revolutionize AI inferencing with metamaterial technology
BitcoinWorld Optical processors breakthrough: Neurophos secures $110M to revolutionize AI inferencing with metamaterial technology AUSTIN, Texas — In a significant development for artificial intelligence infrastructure, photonics startup Neurophos has secured $110 million in Series A funding to commercialize optical processors that could dramatically reduce the energy consumption of AI inferencing. The company’s technology, derived from metamaterials research originally developed for electromagnetic cloaking, promises processing speeds and efficiency gains that challenge Nvidia’s dominance in AI hardware. From invisibility research to optical computing revolution The journey from electromagnetic cloaking to optical processors represents a remarkable evolution in materials science. Two decades ago, Duke University professor David R. Smith pioneered metamaterials research that demonstrated limited invisibility effects at microwave frequencies. Today, that same fundamental research has spawned Neurophos, a startup developing what it calls “metasurface modulators” for optical computing. These modulators function as tensor core processors specifically designed for matrix vector multiplication—the mathematical operation at the heart of AI inferencing. Unlike traditional silicon-based GPUs and TPUs that use electrical signals, Neurophos’s optical processing units (OPUs) manipulate light to perform calculations. This approach offers several theoretical advantages, including reduced heat generation and faster signal propagation. The $110 million funding round and strategic investors Neurophos’s recent funding round attracted significant attention from major technology investors. Gates Frontier, Bill Gates’s venture firm, led the investment, with participation from Microsoft’s M12 fund, Carbon Direct, Aramco Ventures, Bosch Ventures, Tectonic Ventures, and Space Capital. This diverse investor group reflects broad industry recognition of the potential impact of optical computing on AI infrastructure. Dr. Marc Tremblay, corporate vice president and technical fellow of core AI infrastructure at Microsoft, emphasized the technology’s importance in a statement: “Modern AI inference demands monumental amounts of power and compute. We need a breakthrough in compute on par with the leaps we’ve seen in AI models themselves, which is what Neurophos’s technology and high-talent density team is developing.” Technical specifications and competitive advantages Neurophos claims its optical processors achieve remarkable performance metrics compared to current market leaders. According to company data, their chip operates at 56 GHz, delivering 235 Peta Operations Per Second (POPS) while consuming just 675 watts. In contrast, Nvidia’s B200 AI GPU reportedly delivers 9 POPS at 1,000 watts—a significant efficiency gap. Performance Comparison: Neurophos vs. Nvidia Metric Neurophos OPU Nvidia B200 GPU Processing Speed 56 GHz Not specified Peak Performance 235 POPS 9 POPS Power Consumption 675 watts 1,000 watts Energy Efficiency ~0.35 POPS/watt ~0.009 POPS/watt The company’s key innovation lies in the miniaturization of optical components. Traditional optical transistors face significant size limitations, but Neurophos claims its metasurface modulators are approximately 10,000 times smaller than conventional optical components. This miniaturization enables the company to fit thousands of units on a single chip, dramatically increasing computational density. Solving photonic computing’s traditional challenges Photonic computing has long promised advantages over silicon-based electronics, but several persistent challenges have limited commercial adoption. Optical components typically require: Larger physical footprints than silicon transistors Complex data conversion between optical and electronic domains Specialized manufacturing processes incompatible with existing foundries Neurophos addresses these issues through its metasurface technology. The company’s chips can reportedly be manufactured using standard silicon foundry materials, tools, and processes. Additionally, the reduced size of their optical components minimizes the need for frequent data conversion between domains, improving overall efficiency. Dr. Patrick Bowen, CEO and co-founder of Neurophos, explained the technical approach: “When you shrink the optical transistor, you can do way more math in the optics domain before you have to do that conversion back to the electronics domain. If you want to go fast, you have to solve the energy efficiency problem first.” Market timing and competitive landscape Neurophos enters a rapidly evolving AI hardware market dominated by Nvidia, which currently supplies the majority of GPUs powering AI training and inference. However, the company positions its technology as complementary rather than directly competitive, focusing specifically on inference workloads where energy efficiency matters most. The startup acknowledges it faces competition from other photonics companies, though some competitors like Lightmatter have shifted focus to optical interconnects rather than processing units. Neurophos expects its first chips to reach the market by mid-2028, giving the company several years to refine its technology while the AI hardware market continues to expand. Bowen remains confident about the company’s competitive position: “What everyone else is doing, including Nvidia, in terms of the fundamental physics of the silicon, it’s really evolutionary rather than revolutionary. Even if we chart out Nvidia’s improvement in architecture over the years, by the time we come out in 2028, we still have massive advantages.” The environmental imperative for efficient AI The timing of Neurophos’s technology development coincides with growing concerns about AI’s environmental impact. Recent studies indicate that data center energy consumption could double by 2026, driven largely by AI workloads. More efficient processing hardware represents a critical component of sustainable AI development. Neurophos’s optical processors could significantly reduce the carbon footprint of AI inference, which constitutes the majority of computational workload for deployed AI systems. The company claims its technology offers 50x improvements in both energy efficiency and raw speed compared to Nvidia’s current Blackwell architecture. Implementation roadmap and commercial strategy The $110 million funding will support several key initiatives over the coming years. Neurophos plans to develop its first integrated photonic compute system, including datacenter-ready OPU modules, a complete software stack, and early-access developer hardware. The company is also expanding its physical presence with a new engineering site in San Francisco and an expanded headquarters in Austin, Texas. Neurophos has already signed multiple customers, though the company has not disclosed their identities. Microsoft is reportedly “looking very closely” at the startup’s products, suggesting potential integration with Azure’s AI infrastructure. The company’s technology could eventually benefit various applications, including: Large language model inference for chatbots and content generation Computer vision systems for autonomous vehicles and surveillance Scientific computing requiring massive matrix operations Edge AI applications where power constraints are critical Conclusion Neurophos’s $110 million funding round represents a significant vote of confidence in optical computing’s potential to transform AI infrastructure. The company’s metamaterial-based optical processors promise unprecedented efficiency gains for AI inferencing, addressing both performance demands and environmental concerns. While commercial availability remains several years away, the technology could eventually challenge silicon’s dominance in high-performance computing. As AI models grow increasingly complex and energy-intensive, innovations like Neurophos’s optical processors may prove essential for sustainable AI development. FAQs Q1: What are optical processors and how do they differ from traditional chips? Optical processors use light rather than electricity to perform computations. They offer potential advantages in speed and energy efficiency because light generates less heat, travels faster, and is less susceptible to electromagnetic interference than electrical signals. Q2: How does Neurophos’s technology relate to metamaterials research? Neurophos’s optical processors use metasurface modulators derived from metamaterials research originally developed for electromagnetic cloaking. These artificial materials manipulate light in ways natural materials cannot, enabling miniaturized optical computing components. Q3: When will Neurophos’s optical processors be commercially available? The company expects its first chips to reach the market by mid-2028. The current funding will support development of datacenter-ready modules, software stacks, and early-access hardware over the next several years. Q4: How do optical processors address AI’s environmental impact? AI inference consumes substantial energy in data centers. Neurophos claims its optical processors offer 50x improvements in energy efficiency compared to current GPUs, potentially reducing the carbon footprint of AI applications significantly. Q5: What challenges does photonic computing face compared to traditional silicon? Traditional photonic components are larger than silicon transistors, require frequent data conversion between optical and electronic domains, and have faced manufacturing challenges. Neurophos addresses these through miniaturized metasurfaces compatible with standard silicon foundry processes. This post Optical processors breakthrough: Neurophos secures $110M to revolutionize AI inferencing with metamaterial technology first appeared on BitcoinWorld .
22 Jan 2026, 15:12
European AI chief says China’s AI capabilities are closer to the US than many admit
A top European artificial intelligence executive is pushing back hard against the idea that China lags behind America in AI development, calling it nothing more than a “fairy tale.” Arthur Mensch runs Mistral, and he didn’t mince words Thursday at the World Economic Forum in Davos, Switzerland. He said China’s open-source technology capabilities are “probably stressing the CEOs in the US.” That’s not what other tech leaders at Davos were saying. Most of them tried to reassure lawmakers and business people that Chinese AI development sits months or years behind the cutting edge. Google DeepMind’s Demis Hassabis put the gap at about six months for frontier model development. He said Chinese companies haven’t shown they can break new ground. Hassabis did say the gap between China and Western companies might be smaller than people think. He suggested Chinese firms could be just six months behind rather than one or two years. But he stuck to his point that Chinese companies haven’t proven they can push past where things are now. Anthropic’s Dario Amodei went even harder, defending US restrictions on selling advanced tech to China. He compared selling high-end AI chips to the country to “selling nuclear weapons to North Korea.” European challenger eyes enterprise growth Mistral is trying to carve out space in a market where the US and China dominate. Last year, the Paris-based startup pulled in €1.3 billion ($1.5 billion) in investment. ASML Holding, the Dutch chip-machine maker, led the round. It was a rare team-up between two of Europe’s most important tech companies. Mensch said Mistral is going after enterprise clients. Financial companies like HSBC Holdings and BNP Paribas are driving growth. The company wants to top $1 billion in revenue and plans to invest $1 billion in capital spending this year. They’re also looking at acquisition targets. AI becomes a major geopolitical force As Cryptopolitan highlighted earlier, AI has become a big deal in geopolitics. It could reshape economies and how people work in the coming years. Companies and countries are throwing billions at building AI infrastructure and capabilities. Nvidia’s Jensen Huang said Wednesday it would cost trillions. When DeepSeek released its model nearly a year ago, it caused quite a stir. The announcement triggered a stock market drop that temporarily wiped nearly $1 trillion from US and European tech companies. Nvidia lost hundreds of billions in market value. While Mensch has called DeepSeek’s success a win for open-source, the debate over China’s true capabilities continues to divide tech leaders and policymakers on both sides of the Atlantic.’ There’s also a policy shift happening. Trump administration officials are easing restrictions on advanced AI chip exports to China. They’re moving away from policies meant to keep Beijing from accessing American technology for AI development. Sales of the most advanced processors are still blocked for national security reasons, but it’s a big policy change. Advanced AI chips have turned into the new battleground in global tech competition, which is why Amodei’s warning against sales to rival nations got attention. The back-and-forth shows how much disagreement exists among industry leaders about where China really stands and what should be done about its growing technological strength. China has been making moves to continue AI innovation while universities introduce DeepSeek-based courses, signaling the country’s commitment to advancing in the field. If you're reading this, you’re already ahead. Stay there with our newsletter .
22 Jan 2026, 15:01
Julie Sweet calls data centers a critical national asset for AI growth at Davos
Accenture CEO Julie Sweet today called on governments worldwide to place data center infrastructure at the core of their national artificial intelligence (AI) strategies, stressing that computational capacity must keep pace with the rapid pace of AI innovation. Speaking on the sidelines of the World Economic Forum’s annual Davos meeting, Sweet highlighted that data centers, the physical backbone where AI systems compute and store data, are no longer a technical afterthought but a critical national asset. She told reporters that if countries want to harness the transformative power of AI, they must treat data centers as central infrastructure , like roads or power grids. “Data centers shouldn’t just be an afterthought in national AI plans,” Sweet told delegates and media. “They are the foundation of secure, scalable, and sovereign AI deployment that will drive future economic growth and protect data privacy.” Accenture Research has been training staff in new tech skills for years now Earlier this month, Accenture struck a deal to buy Faculty, a British artificial intelligence company, as it seeks to position itself as a technology frontrunner. The firm has also been retraining staff in new technologies while phasing out employees who fail to adapt since 2019. Sweet shared , “Technology is a basic skill now. And once you start making that investment, people get less scared about their jobs because they know you’re invested in making sure they can do the next jobs.” Sweet also pointed out that designing new entry-level jobs requires educational support and cooperation with schools and universities, and that such efforts are still in the initial stages. Her statements come in the wake of Nvidia Jensen Huang’s Davos comments that the AI surge will allow plumbers, electricians, and construction workers to earn six-figure incomes building data centers. Sweet also said that the critical error for CEOs is treating AI as a target in its own right rather than aligning it with their company’s goals, noting that their business strategy should come first. Sweet noted that business leaders are seeing AI as more than just a cost-cutting tool Sweet also told reporters on the sidelines of the World Economic Forum in Davos that an increasing number of executives now view AI as a growth driver rather than merely a cost-reduction tool, adding that she is optimistic about agentic commerce and other AI applications. A Pulse of Change survey by Accenture Research found that most leaders see AI as a tool for growth, with revenue gains outweighing its cost-saving potential. She asserted, “Companies are led by humans, and they will win by tapping into human creativity, adding that the best AI future would be using technology as a tool rather than relegating people to a supporting role. Sweet argued that being a “human in the loop” isn’t particularly inspiring for people. In Accenture’s survey, 83% of non-executive employees believe their companies would continue investing in AI in ways that benefit both staff and business results, regardless of an AI downturn. However, the survey found that only one in five respondents feels they are actively helping shape how AI changes work. Even fewer—17%—said they enjoy using it and exploring new applications. Sweet also acknowledged that while many employees use AI in their personal lives, there is still considerable anxiety about it in the workplace. Late last year, the Accenture executive had insisted that it would take at least a few years for most companies to move beyond the slow, hard phase of artificial intelligence. At the time, she explained that most business leaders and processes need to change to bolster the technology. Get seen where it counts. Advertise in Cryptopolitan Research and reach crypto’s sharpest investors and builders.
22 Jan 2026, 14:59
Tron Founder Justin Sun Invests $8M in River’s Stablecoin Abstraction Technology
Tron founder Justin Sun invested $8 million in DeFi project River to support ecosystem integration on the Tron blockchain and deployment of River’s chain abstraction stablecoin infrastructure. The deal positions Tron to leverage River’s cross-chain technology through satUSD, a stablecoin mintable at a 1:1 ratio with USDT, USDD, or USD1. River announced the funding on X, emphasizing its mission to build a system that connects every asset to its opportunity while allowing value to flow freely across ecosystems without locking capital away. $8M Strategic investment by @justinsuntron This investment supports ecosystem integration on @trondao and the deployment of River’s chain abstraction stablecoin infrastructure. River connects cross ecosystem assets and liquidity into TRON through satUSD, which can be minted 1… pic.twitter.com/3t9P069tPI — River (@RiverdotInc) January 21, 2026 The investment comes weeks after MaelstromFund, founded by BitMEX co-founder Arthur Hayes, also backed the project in early January. River Bags Stablecoin Integration Across the Tron Ecosystem Per the announcement, Justin Sun’s capital will support multiple deployments, including stablecoin pools alongside USDT and USDD on SUN, lending and borrowing on JustLend, and price feeds provided by WinkLink. Integration extends across core assets,s including USDT, TRX, wBTC, BTT, JST, SUN, WIN, and NFT use cases, with native sTRX staking yield serving as the initial entry point. River also plans to launch Smart Vault and Prime Vault products targeting yield strategies for stablecoins, TRX, and other core Tron assets. Since the funding announcement, River’s ($RIVER) token appreciated over 20%, reaching an all-time high of $48.74. The token posted over 800% gains in the last 30 days to reach a market capitalization of around $840 million, jumping from $8 to the current $42.68 after starting January with approximately $100 million market cap. Source: Coingecko Hayes’ Maelstrom investment in early January triggered a 600% surge for RIVER within weeks, with the token rising from around $3 to $19. Market observers attributed the rally to Hayes’ endorsement and his stated belief in chain abstraction technology as fundamental to DeFi’s next growth phase. River currently integrates with over 30 protocols across major ecosystems, including Ethereum, BNB Chain, and Base, with satUSD circulation exceeding $100 million. Legal Challenges Shadow Sun’s Investment Activity Sun’s recent capital commitment unfolds amid ongoing legal scrutiny around the alleged misappropriation of TrueUSD (TUSD) stablecoin reserves. Last November, a judge at the Dubai International Financial Centre imposed a worldwide freeze on $456 million in assets tied to TUSD reserves, linked to Sun’s earlier bailout of the token. According to case filings , Techteryx, which acquired TrueUSD in 2020, failed to redeem a large portion of its U.S. dollar reserves managed by First Digital Trust between 2022 and 2023. Counsel for Techteryx stated that reserves originally custodied in Hong Kong saw around $468 million invested in the Aria Commodity Finance Fund, though nearly $456 million was transferred directly to Aria Commodities DMCC. The diverted funds gave rise to claims of breach of trust and knowing receipt, prompting the proprietary injunction and subsequent global asset freeze. Beyond Dubai, Congressional Democrats on January 15 formally accused the Securities and Exchange Commission of operating a pay-to-play scheme in its handling of crypto enforcement cases, with particular focus on the agency’s treatment of Sun. Representative Maxine Waters sent a detailed letter to SEC Chairman Paul Atkins highlighting Sun’s extensive financial relationship with Trump family ventures, noting his $75 million investment in World Liberty Financial. Sun is also a top holder of Trump’s memecoin , which earned him an invitation to a May 2025 White House dinner for major investors. Tron founder @justinsuntron has received a Trump-branded Golden Tourbillon watch for being the top holder of President @realDonaldTrump ’s memecoin. #Trump #Sun https://t.co/NI4bVy3smJ — Cryptonews.com (@cryptonews) May 23, 2025 Regulators also claimed Sun engineered the offer and sale of two crypto asset securities without proper registration while directing hundreds of thousands of TRX wash trades that generated approximately $31 million from unsuspecting investors. Judge Vernon Broderick of the Southern District of New York sustained core allegations in a parallel private class action, finding that plaintiffs plausibly alleged Sun and Tron illegally sold TRX as an unregistered security. Despite these ongoing legal challenges, Sun continues to expand his cryptocurrency portfolio and investments, with Bloomberg estimating his net worth at approximately $12.5 billion. The post Tron Founder Justin Sun Invests $8M in River’s Stablecoin Abstraction Technology appeared first on Cryptonews .
22 Jan 2026, 14:11
ETHDenver Returns for its Ninth Edition, Driving Web3’s Global Agenda for 2026
Denver, CO, January 20, 2026 – ETHDenver , the world’s largest Web3-focused innovation festival, returns for its ninth edition from February 17–21, 2026. Since its founding in 2018, ETHDenver has generated nearly $250 million in cumulative economic impact for Colorado and has served as a launch point for thousands of open-source projects and startups worldwide. Building on the momentum of 2025 and the continued growth of its global community, ETHDenver’s organizers have curated full and half-day content blocks across the main event. These blocks feature back-to-back sessions aligned to specific tracks and communities, creating clearer pathways for attendees to engage deeply with the topics and builders most relevant to them. For 2026, ETHDenver will take place on a brand-new, state-of-the-art campus-style venue designed to support focused programming, seamless movement between sessions, and integrated onsite side events that keep the community connected throughout the week. This year’s summit-style format reflects ETHDenver’s expanding role as a convening ground for the industry’s most influential operators, capital allocators, and public-sector leaders. “ ETHDenver has always been about meeting builders where they are ,” said John Paller , Founder and Chief Steward of ETHDenver. “ When side events doubled because our community craved focused experiences, we evolved. The summit format gives every attendee curated pathways to the conversations that matter most to them .” For nearly a decade, ETHDenver has functioned as a builder-led platform supporting open-source infrastructure, startups, and public-sector engagement worldwide. “ The culture that started in hacker basements is now influencing national policy and boardroom strategy ,” Paller said. “ We did not move toward the world. The world moved toward us.” Blockchain technology is no longer an experiment but a global industry, and the United States is increasingly setting its pace and direction. At the center of that momentum is ETHDenver, where more than 25,000 builders, founders, institutions, and policymakers from 140 countries and all 50 U.S. states attend with the shared goal of shaping Web3’s direction. ETHDenver operates as a community-owned organization run by SporkDAO , uses a patronage model that shares profits directly with the members who build and sustain the event. SporkDAO distributed more than $470,000 in community-generated profits from ETHDenver across two distributions: $52,800 in December 2024 and $420,000 in October 2025. The October distribution rewarded approximately 7,500 qualifying members based on their staked $SPORK, demonstrating a model where contributors directly participate in the value they help create. ETHDenver 2026 offers opportunities for both first-time attendees and seasoned builders to explore, build, and contribute at the heart of a truly community-driven ecosystem where there is somETHing for everyone. About John Paller John Paller is a Denver-based blockchain entrepreneur and futurist who founded ETHDenver , the world’s largest Web3 and Ethereum-focused hackathon and festival, attracting more than 100,000 participants across eight seasons. He also founded Opolis , an innovative employment ecosystem for independent workers. With over 20 years of experience in talent acquisition and HR before discovering Ethereum in 2014, Paller now serves on the SporkDAO Board , where he pioneers patronage token models to sustainably fund decentralized ecosystems. His work bridges traditional employment systems with decentralized innovation and has earned him recognition as one of Denver Business Journal’s “ 40 Under 40 ” (2014). Media Contact Emma Murphy [email protected] The post ETHDenver Returns for its Ninth Edition, Driving Web3’s Global Agenda for 2026 appeared first on Cryptonews .
22 Jan 2026, 14:05
Coinbase names Stanford and Harvard experts in quantum assessment board
Coinbase has announced it has assembled a team of experts to assess the emerging threat of quantum computing to cryptography. The exchange has recruited experts in computer science, cryptography, and fintech from Harvard and Stanford. Coinbase, a U.S.-based crypto exchange, has announced the creation of a task force composed of experts in computer science, cryptography, and fintech from Stanford, Harvard, and the University of California. The team will investigate the potential risks posed by quantum computing to cryptography. The team will also involve experts from Coinbase, the Ethereum Foundation, and the DeFi platform EigenLayer. Coinbase’s official says quantum computing could decrypt wallet private keys Coinbase Chief Information Security Officer Jeff Lunglhofer said in an interview that quantum computing could defeat current encryption mechanisms, including those designed to protect wallets and private keys where users store their digital assets. He emphasized that the underlying technology around crypto assets relies on complex mathematical problems that a regular computer would take “thousands of years to solve.” Bitcoin uses private keys to secure the network. These private keys are long strings of random numbers and letters that are visible only to their owners and can be guessed only through a long series of trial-and-error attempts. However, quantum computing has “a million times the horsepower,” enabling it to solve these problems more quickly. Lunglhofer said the security risk exists. However, he noted that quantum computing is not an immediate threat to cryptography, and this may remain so for the next decade. The Coinbase exec’s opinion aligns with that of other professionals, who say that large tech companies like Google have spearheaded the development of quantum computers and their capabilities over the last decade. However, their developments operate only at a small scale and lack advanced capabilities to decrypt the security mechanisms in the cryptocurrency ecosystem that protect Bitcoin and other blockchain networks. Lunglhofer also highlighted that Coinbase’s purpose for the new advisory board is to investigate the imminent impact of quantum computers in a “non-hype-based way,” to promote future security efforts. He added that when the quantum computing era begins, the industry will employ safety measures, including using larger private keys and using what he called “noise” to prevent computers from detecting the location of the keys. However, he said that incorporating these security measures will take years to implement. David Duong says CRQCs are the real threat to blockchain technology Global Head of Investment Research at Coinbase, David Duong, also explained the quantum threat to cryptography in a LinkedIn post . He also noted that the threat is not immediate, but highlighted growing concerns among crypto investors about technological advancements. The Coinbase official said that quantum computers will solve “some of the world’s most complex problems,” but they will also require upgrades to many of the cryptographic systems currently in use. According to Duong, the ultimate risk will emerge at what he called “Q-day,” when cryptographically relevant quantum computers (CRQCs) will have the capacity to run Shor’s and Grover’s algorithms, jeopardizing Bitcoin’s underlying cryptographic technology. Duong said that quantum computing poses two threats to cryptography, including the Elliptic Curve Digital Signature Algorithm (ECDSA) for transaction signatures and SHA-256 for proof-of-work mining. He added that a breakthrough in quantum computing could create loopholes that attackers can exploit to breach private keys and steal from vulnerable wallets. Duong also mentioned that advanced quantum computers could mine blocks more efficiently, disrupting Bitcoin’s economic and security model. The smartest crypto minds already read our newsletter. Want in? Join them .
