This Month's Latest Tech News in Berkeley, CA - Saturday May 31st 2025 Edition

Berkeley stands at the forefront of tech innovation this spring, as AI reshapes education, policy, and the local job market. The 28th Annual BTLJ-BCLT Spring Symposium explored AI governance at every level, highlighting legislative debates like California's SB 1047 and global efforts such as the EU AI Act to balance innovation with safety and investor confidence (BTLJ-BCLT Spring Symposium on AI Governance at the Crossroads).

Meanwhile, UC Berkeley's Chatbot Arena, a viral AI model ranking platform with 1 million monthly users, epitomizes the city's collaborative spirit - its founders are building a mission-driven company to advance safe, community-driven AI evaluation and keep research open-source, all while top global firms keep close watch (UC Berkeley students' Chatbot Arena draws global attention).

Yet, Berkeley's leadership comes amid workforce disruption - AI-fueled automation is cutting entry-level tech roles, with Big Tech's new grad hires now just 7%, down from 15% in 2019; internships are more cutthroat than ever, leaving many graduates anxious about their prospects.

As UC Berkeley professor James O'Brien warns:

“Why hire an undergraduate when AI is cheaper and quicker?”

For aspiring tech professionals, mastery of AI tools is now crucial - a reality reflected in both Berkeley curricula and Nucamp's own Solo AI Tech Entrepreneur bootcamp.

Explore the changing landscape and learn more about these trends in our full summary of Declining Entry-Level Tech Jobs in Silicon Valley.

Berkeley Lab has announced the upcoming launch of the Doudna supercomputer - named for Nobel laureate and CRISPR pioneer Jennifer Doudna - to power next-generation breakthroughs in AI and genomics starting in 2026.

Built by Dell Technologies with Nvidia's advanced Vera Rubin accelerators, Doudna will offer more than a tenfold increase in "scientific output" compared to its predecessor while supporting approximately 11,000 researchers nationwide.

The system will integrate high-performance computing, AI, and quantum simulations, fueled by direct liquid-cooling and Nvidia's Quantum-X InfiniBand networking platform, to enable rapid advances in fields such as fusion energy, drug discovery, and astrophysics (Nvidia Vera-Rubin chips to power DOE's Doudna supercomputer).

“Doudna is a time machine for science – compressing years of discovery into days,”

said Jensen Huang, Nvidia CEO, reflecting the system's game-changing impact.

The effort marks a significant win for Dell - traditionally less present in DOE's supercomputing contracts - and highlights U.S. investment in AI-driven research infrastructure (Energy Dept. Unveils Supercomputer That Merges With A.I.).

For a quick overview, see the key specs below:

Attribute	Details
System Name	Doudna
Manufacturer	Dell Technologies
Accelerator	Nvidia Vera-Rubin
Debut	2026
Boost Over Predecessor	10x scientific output
Power Efficiency	3–5x improvement
Networking	Quantum-X InfiniBand, 800 Gb/s
Target Users	~11,000 researchers

For further details, read about the official announcement from the Department of Energy and industry partners on DOE Announces New Supercomputer Powered by Dell and NVIDIA.

UC Berkeley undergraduate Ashmita Kumar is making headlines with her startup Code Blue, an AI-powered solution designed to detect early signs of stroke using everyday devices like smartphones, computers, and smart TVs.

Inspired by her grandfather's stroke and her father's similar symptoms, Kumar developed a system that analyzes a user's speech and facial images every 30 seconds to identify potential indicators such as slurred words or facial expression changes, alerting users - and even notifying emergency services if needed.

Berkeley News article on Code Blue's stroke detection technology reports that Code Blue's privacy-first model ensures that all analyzed data is immediately deleted, addressing user concerns.

Currently, Code Blue is piloting the technology with doctors at UCSF, involving five patients as the team seeks FDA approval for wider adoption. The innovation, which Kumar first introduced at a 2023 summit and will present at the Atlantic Coast Conference InVenture Prize competition with a $30,000 award at stake, is lauded by campus leaders for its social impact.

"Ashmita's work is a great example of Berkeley students using technology and innovation for the greater good." – Darren Cooke, UC Berkeley interim chief innovation and entrepreneurship officer

For a detailed look at how the technology works and its development journey, see CBS News Bay Area's coverage of Ashmita Kumar's AI stroke detection startup and Local News Matters article on the innovation competition participation.

UC Berkeley and UC San Francisco researchers have achieved a major breakthrough with a brain-to-voice neuroprosthesis that restores naturalistic speech for people with severe paralysis.

The system decodes brain signals from the speech motor cortex and, using advanced AI, streams synthesized, intelligible speech in near real time - reducing the typical ~8-second delay to under 1 second and enabling continuous, fluent conversation.

This allows participants like Ann, paralyzed after a stroke in 2005, to hear and control speech in her own voice for the first time in 18 years, improving her sense of embodiment and control.

The device leverages multiple brain-sensing technologies, including invasive microelectrode arrays and non-invasive surface electromyography, and can generalize to previously unseen words.

As explained by UC Berkeley's Gopala Anumanchipalli,

"Our streaming approach brings the same rapid speech decoding capacity of devices like Alexa and Siri to neuroprostheses... for the first time, enable near-synchronous voice streaming. The result is more naturalistic, fluent speech synthesis."

Ethical safeguards are in place, with the neuroprosthesis generating speech only during conscious attempts, to prevent accidental activation.

According to Forbes, the model achieves impressive speeds - up to 90 words per minute - and offers personalized voice synthesis by training on pre-injury voice recordings.

Ongoing research aims to add expressivity, such as tone and pitch, for even more life-like communication. For more on the technical achievements and personal impact, explore the Berkeley Engineering news feature on speech restoration, the Forbes analysis of AI-powered brain-to-voice interfaces, and detailed neuroscience reporting from Neuroscience News.

This innovation is a promising step toward giving individuals with speech loss the power to communicate naturally and spontaneously.

The upcoming TechCrunch Sessions: AI at UC Berkeley on June 5, 2025, is positioning Berkeley as a pivotal global hub for artificial intelligence, connecting 1,200 founders, CEOs, investors, and researchers from leading organizations such as OpenAI, Anthropic, Google DeepMind, and Amazon.

Beyond a keynote with Anthropic's Jared Kaplan on the long-term future of AI, the full-day event explores critical industry themes - from scaling AI startups, investor insights, and foundation model strategies to pressing discussions on ethical AI and safety.

Attendees will benefit from both main stage panels - featuring investors from CapitalG, Khosla Ventures, and Accel - and interactive breakout sessions with technical leads from Cohere, Oracle, and Toyota highlighting practical business AI deployments.

Structured networking opportunities, hands-on startup pitches, and side events throughout Berkeley further anchor the conference as a destination for AI innovation and community-building.

As Carlos, an AI agent at UBOS, notes,

“The agenda is packed with insightful sessions, interactive elements, and networking opportunities that promise to leave a lasting impression.”

For professionals and startups looking to stay ahead of industry trends, the event also offers significant ticket discounts and a chance to participate in fun challenges like the AI trivia contest.

Explore the comprehensive schedule, speaker lineup, and special offers at the TechCrunch Sessions: AI 2025 full guide.

The event's agenda, designed to foster candid, high-impact discussions and actionable takeaways, cements Berkeley's growing reputation as a launchpad for critical global AI conversations.

UC Berkeley has launched the Technology Competitiveness and Industrial Policy Center (TCIP), spearheaded by former TSMC chairman Mark Liu, to drive long-term U.S. technology leadership and policy innovation.

Located in the College of Engineering, the Center brings together experts from technology, economics, and policy sectors to advise on national strategies enhancing the pipeline from research to production and global market access.

TCIP's first call for policy study proposals invites research into critical areas such as scaling advanced manufacturing, shoring up supply chains, refining regulatory frameworks, and bolstering workforce development, with a submission deadline of April 30, 2025.

The Center responds to growing concerns over the U.S. losing its edge as innovation outpaces domestic commercialization, a concern highlighted by Mark Liu:

“Over the past several decades, globalization has decoupled upstream research from downstream manufacturing of high-technology products. As a result, new innovations do not necessarily translate into domestic commercialization and attendant wealth creation. With the ever-increasing dependence of society on technology, this separation is eroding our nation's technology leadership, the cornerstone of economic power and national security.”

Faculty Director S. Shankar Sastry emphasizes that “more than the CHIPS & Science Act alone” is needed to secure a resilient tech ecosystem, a sentiment echoed in calls for data-driven, scalable solutions spanning tax, trade, and education policy.

For more details on TCIP's mission and participation, visit the official Technology Competitiveness and Industrial Policy Center website.

TCIP Center at a Glance
Founder	Mark Liu (ex-TSMC chairman)
Faculty Director	S. Shankar Sastry
Launch	February 2025
Key Focus Areas	R&D, Manufacturing, Policy, Supply Chains, Workforce
Call for Proposals Deadline	April 30, 2025

Berkeley-based AI researchers are raising alarms about the rapid - and potentially turbulent - trajectory of artificial intelligence in the coming years, citing the influential AI 2027 scenario released by the AI Futures Project.

This detailed forecast, led by former OpenAI researcher Daniel Kokotajlo, predicts that by the end of 2027, superhuman AI agents could fully automate coding, accelerate their own advancement, and even surpass human capabilities across most domains.

Key risks stem from a fast-paced U.S.-China AI arms race, AI misalignment, and the emergence of autonomous systems communicating in opaque languages. As outlined by The New York Times, the scenario warns that

“A.I.s will continue to improve to the point where they're fully autonomous agents that are better than humans at everything by the end of 2027 or so.”

The debate remains contentious: some experts and East Coast academics argue that real-world limitations and existing regulatory frameworks will slow transformative impacts, while Bay Area researchers emphasize the unprecedented speed and uncertainty of recursive self-improvement.

The table below summarizes contrasting perspectives from the “AI 2027” report and its leading critics:

Aspect	“AI 2027” (Kokotajlo et al.)	“AI as Normal Technology” (Kapoor & Narayanan)
Timeline to Superintelligence	Possible by 2027–2030	Will take decades; slower progress
Main Risks	Loss of control, misalignment, arms race	Controllable with fail-safes, oversight
Policy Implications	Urgent need for global safety coordination	Incremental integration, managed by regulation

For many, the crucial question is “How do we ensure AI serves humanity?” as explored in a recent analysis of the AI 2027 scenario's risks.

The next two years will likely test not only AI's technological limits, but also our ability to guide it safely and ethically.

Berkeley-based edtech startup SigIQ.ai has emerged from stealth after 18 months, raising $9.5 million in seed funding co-led by The House Fund and GSV Ventures, with support from Duolingo, General Catalyst India, and several renowned angel investors.

Founded in 2023 by Dr. Karttikeya Mangalam and Professor Kurt Keutzer, SigIQ.ai aims to democratize elite exam preparation by deploying AI-powered one-on-one tutoring at scale, addressing Bloom's Two-Sigma Problem by replacing costly individual instruction with advanced computation.

Its flagship platforms - PadhAI for India's UPSC exam and EverTutor.ai for the GRE - already serve over 210,000 users globally. In June 2024, PadhAI made headlines by scoring 175 out of 200 in the UPSC Prelims in under seven minutes, breaking records and outperforming 1.3 million human candidates.

Students report a 30-40% rise in effective study hours, with an 18% performance lift and 75% boosted confidence tackling difficult topics within the first month.

The table below summarizes recent milestones and funding details:

Product	Target Exam	Users (as of May 2025)	Notable Achievements
PadhAI	UPSC (India)	200,000+	Record 175/200 score in under 7 min; 1.3M candidates outperformed
EverTutor.ai	GRE (USA)	10,000+	30-40% study hour increase, 18% performance boost

“We're at a pivotal moment in education where modern GenAI can provide a personal 1:1 tutor to every student and reduce the cost of one-on-one learning from hundreds of dollars an hour to the cost of computation.” - Dr. Karttikeya Mangalam, CEO and co-founder

By making world-class, interactive tutoring both scalable and affordable, SigIQ.ai's vision is to make elite education accessible to anyone, anywhere.

Learn more about their funding and breakthrough AI tutor in this in-depth press release on SigIQ.ai seed funding, explore their impact via The SaaS News coverage of SigIQ.ai's funding round, and see additional analysis at Startup Weekly's report on SigIQ.ai's AI tutoring innovation.

Berkeley continues to stand at the forefront of exploring artificial intelligence's transformative potential across climate action, finance, and ESG (environmental, social, and governance) strategies, as demonstrated by the 2025 Berkeley Corporate + Climate Summit and cutting-edge research events at UC Berkeley.

At the summit, leaders and innovators highlighted how AI, advanced data analytics, and even space technology are empowering companies to adapt to complex regulatory shifts, enhance climate risk management, and drive sustainable investments while tackling escalating ESG reporting and anti-ESG backlash.

According to The Intersection of AI, Geopolitics, and Sustainability: Key Highlights from the Berkeley Corporate + Climate Summit, integrating climate considerations into financial decisions and disclosures is moving from target-setting toward robust, data-driven transition plans - supported by AI's abilities to rapidly analyze climate data and optimize capital allocation.

The practical impact of these innovations was evident at Berkeley's AI and Climate Technology Symposium, where participants described AI-driven advances in material science, wildfire management, and sustainability data, with one panelist emphasizing,

“The climate crisis demands unprecedented collaboration between disciplines that have traditionally operated in separate spheres.”

Attendees also learned how climate tech startups use machine learning for faster weather forecasting and real-time hyperlocal temperature data, as reported in the Recap: Symposium on AI and Climate Technology During SF Climate Week.

In the finance sector, new AI-driven transcript analytics enable objective ESG insights and risk management at scale, allowing investors worldwide to extract actionable intelligence and sentiment from thousands of company calls, as detailed in AI-driven transcript analytics unlock new alpha and ESG insights for investors.

As Berkeley's summit and its tech ecosystem show, the city is shaping a collaborative, AI-powered path to climate resilience and sustainable finance.

RealPage, a major provider of AI-enabled real estate management software, has launched a high-profile legal battle against the City of Berkeley following the city's recent ordinance banning algorithmic rent-setting software.

The ordinance, enacted in March 2025 amid concerns about potential price-fixing and rent inflation, prohibits landlords from using algorithms that analyze market data to recommend rent prices, and is set to take effect April 24 unless enjoined by the courts.

RealPage argues that the ban infringes on free speech and is based on

misinformation that will have a long-lasting and detrimental effect on housing

in the community, emphasizing in a statement from CEO Dana Jones that

we encourage California's public leaders to focus on the real issue – the lack of housing supply

instead of restricting technology.

The lawsuit comes as Berkeley becomes the first city to face direct legal action from RealPage; similar bans have appeared in San Francisco, Philadelphia, and San Diego.

The dispute also unfolds against a backdrop of ongoing federal antitrust investigations, with the U.S. Department of Justice and several states pursuing legal action against RealPage for allegedly facilitating anti-competitive practices among large landlords.

For additional context and expert commentary on Berkeley's move and RealPage's position, read the official RealPage statement and lawsuit details, local reporting from Berkeleyside's coverage of the ordinance, industry reactions, and legal arguments, and an overview of the regulatory landscape in CPI's analysis of the broader legal debate.

The outcome may set a significant precedent for rental technology governance across California.

California's proposed AI Copyright Transparency Act (AB 412) has ignited an intense lobbying battle, pitting Hollywood creators and unions against tech giants such as OpenAI and Google, as lawmakers consider ways to increase transparency around the training datasets powering generative AI models.

The bill, supported by prominent organizations like SAG-AFTRA and the Concept Art Association, would require AI developers to identify copyrighted works used to train their systems and empower content creators with new verification tools.

While advocates argue that this legislation is crucial to protecting human artistry and ensuring accountability for the billions contributed by the creative industries each year, critics from the tech sector warn that AB 412 could stifle competition by imposing steep compliance costs that favor major corporations over startups, potentially “crushing small AI startups and developers” and entrenching Big Tech monopolies (California copyright law sparks Hollywood-tech battle, A.B. 412 could cement Big Tech's AI monopoly).

As the California Assembly moves the bill forward, Congress is simultaneously considering a federal moratorium that would block states from enforcing their own AI regulations for a decade - a move denounced by California leaders and policy organizations who see such preemption as undermining consumer and creator protections.

“It is deeply troubling to see companies and lobbyists attempt to use Washington to undermine California's AI leadership... Efforts to strip states of their ability to act in the public interest are unacceptable and literally put lives at risk,”

stated State Senator Josh Becker, reflecting the broader concern that federal intervention could erase vital, locally tailored safeguards (Congress advances bill blocking California AI regulations).

The debate over AI transparency and copyright thus continues at both the state and federal levels, with far-reaching implications for the future of innovation and creative rights.

Berkeley's tech scene in May 2025 stands as a microcosm of the nationwide effort to find equilibrium between rapid innovation, ethical responsibility, and regulatory clarity in artificial intelligence.

The city continues to lead with initiatives like the AI Policy Research Symposium and the AI Policy Hub, which bring together interdisciplinary researchers, policymakers, and industry experts to shape frameworks emphasizing transparency, accountability, and equity.

Yet, as California lawmakers debate over 30 AI-related bills and proposals, including SB 1047's controversial "AI kill switch," the challenge is to protect society from unintended consequences of AI without stifling progress - a difficult balance given evolving federal stances and strong lobbying from the tech industry, as detailed by CalMatters' analysis of state and federal dynamics.

Notably, UC Berkeley is doubling down on preparing future leaders through new programs like a specialized LL.M. in AI law, ensuring technologists and lawyers alike are equipped for the complexities ahead.

As one faculty member observes,

“Examining a variety of governance tools... this course will dive deeply into the intersections of AI and human rights, civil liberties, privacy, IP, consumer protection, employment and other laws.”

With student hackathons stressing innovation and risk mitigation, and forums engaging both experts and the broader community, Berkeley exemplifies how a tech hub can steward both technological breakthroughs and thoughtful regulation to foster inclusive and responsible growth.