Hello and welcome to Sync #487!

The big news this week was OpenAI’s massive $6.6 billion funding round, which brought the company’s value to $157 billion. We will take a closer look at this deal, piecing together reports from various sources to understand what it means for OpenAI and its investors.

In other news, California Governor Gavin Newsom vetoed the controversial AI bill SB 1047. Elsewhere in AI, DeepMind open-sourced an AI model that designs computer chips, Snap introduced AI video tools to Snapchat, and a popular tech YouTuber shared his experiences of having his voice cloned.

In robotics, a new humanoid robot has been revealed, and roboticists demonstrated an intriguing robot that can detach its hand.

We will finish this week’s issue with a new episode in the fight over the CRISPR patent, some ideas on improving drug development, and a breakthrough map of a fly’s brain.

Enjoy!

OpenAI raises over $6.6 billion

After weeks of rumours and speculation, it has happened. OpenAI has closed one of the largest VC funding rounds ever, raising $6.6 billion, which values the company at $157 billion. This deal makes OpenAI one of the highest-valued privately held companies globally, alongside SpaceX and TikTok, and worth more than 87% of S&P 500 companies.

This funding round was led by Thrive Capital, which previously invested in OpenAI. Other participants included Microsoft, Nvidia, SoftBank, Khosla Ventures, Altimeter Capital, Fidelity, and MGX, a Abu Dhabi-based tech investment firm. Notably absent from this list is Apple, which was reportedly in the talks to invest into OpenAI.

Interestingly, the deal comes with some strings attached, both for OpenAI and for the investors. For OpenAI, the deal might mean a transition from its unusual nonprofit structure to a for-profit company which is more suitable for investors. Meanwhile, as Financial Times reports, investors were asked to not invest in OpenAI competitors, such as Anthropic or xAI. If the reports of exclusivity are true, then that could have a big impact on the future investments in AI space.

In addition to $6.6 billion raised, OpenAI also secured a $4 billion credit line from nine banks—JPMorgan Chase, Citi, Goldman Sachs, Morgan Stanley, Santander, Wells Fargo, SMBC, UBS, and HSBC. “This credit facility further strengthens our balance sheet and provides flexibility to seize future growth opportunities,” said Sarah Friar, CFO of OpenAI, in an announcement.

What will OpenAI use all this money for?

In total, OpenAI has secured over $10 billion in funding. So, what will OpenAI use all this money for now?

According to OpenAI, the new funding will allow the company to “double down“ on its leadership in frontier AI research, increase compute capacity and build new AI tools.

In practice, it means OpenAI will use these funds to develop and train new models and add new functionalities, like the recently announced Canvas coming soon to ChatGPT (a clear response to Anthropic’s Artifacts) or the new features available for developers announced during this year’s first Dev Day.

Some of those funds will go towards securing deals with copyright holders to use their data in OpenAI products. In the last year, OpenAI signed content deals with multiple partners, including Condé Nast, Time, The Atlantic, Vox Media, News Corp, Financial Times, with possibly more to come. Additionally, OpenAI is involved in multiple lawsuits, with the largest of them, filed by The New York Times, still unresolved and potentially costing OpenAI billions of dollars even if both companies opt for a settlement.

There is also a possibility of OpenAI trying to design its own AI chips and build its own data centres. Currently, OpenAI heavily relies on Nvidia and Microsoft for these needs. Although designing custom chips is an expensive project, it can pay off in the long term, as chips specifically designed for a particular operation are more efficient than the general-purpose chips Nvidia offers. Additionally, this would reduce OpenAI's dependency on Nvidia.

OpenAI buys itself time to become profitable and to become the next tech giant

But more importantly for OpenAI, the $10 billion it secured buys the company time to become profitable. Earlier this year, in the summer, The Information reported that OpenAI is bleeding money. According to their report, OpenAI has spent roughly $7 billion on model training and $1.5 billion on staffing, and is on track to lose $5 billion this year.

Although OpenAI is burning through cash right now, the company projects to turn this trend around. OpenAI’s revenue is projected to reach $3.4 billion in 2024, $2.7 billion of which could come from ChatGPT alone and its 11 million paid subscribers. Next year, the company is expected to generate over $10 billion in revenue in 2025 and reach $100 billion in revenue by 2029.

OpenAI is at an interesting point right now. The company is the poster child of the current AI boom, which it kickstarted with the release of ChatGPT in November 2022. Since then, it has grown from a relatively small AI lab into one of the most valued startups in the world, with some viewing OpenAI as the next tech giant, akin to Google and Facebook.

However, that transformation did not come easily and last year was turbulent for OpenAI. Sam Altman was ousted as a CEO to then come back to his role a couple of days later. The release of GPT-4o was tainted by the controversy’s surrounding ChatGPT’s voice that was very familiar to Scarlett Johansson’s voice. Whistleblowers raised concerns over the safety culture inside the company. The company started become more secretive and stopped sharing details of its models, prompting many to question the validity of the word “open” in OpenAI. Now there are talks about OpenAI becoming a for-profit company, which will essentially end the original vision for the company as an open and independent counterweight to other AI labs like DeepMind.

Additionally, many high-profile people have left the company. Recently, Mira Murati, OpenAI's CTO, announced her departure from the company, along with chief research officer Bob McGrew and research VP Barret Zoph. They join a growing list of high-profile departures over the last year, including Andrej Karpathy (a prominent AI researcher), Ilya Sutskever (OpenAI co-founder and former chief scientist), Jan Leike (former leader of the Superalignment team), and John Schulman (one of OpenAI's co-founders). Additionally, Greg Brockman, the company's president, is currently on sabbatical. Out of fifthteen people who founded OpenAI in 2015, only three remain.

We will have to see what the future holds for OpenAI. Will it deliver on its overarching goal of building AGI to benefit all of humanity, or will it collapse under its own weight?

Leave a comment

If you enjoy this post, please click the ❤️ button or share it.

Share

Do you like my work? Consider becoming a paying subscriber to support it

Become a paid subscriber

For those who prefer to make a one-off donation, you can 'buy me a coffee' via Ko-fi. Every coffee bought is a generous support towards the work put into this newsletter.

Your support, in any form, is deeply appreciated and goes a long way in keeping this newsletter alive and thriving.

🧠 Artificial Intelligence

Gov. Newsom vetoes California’s controversial AI bill, SB 1047
California Governor Gavin Newsom vetoed SB 1047, a controversial bill aimed at regulating AI development by holding companies accountable for implementing safety protocols. The bill, authored by State Senator Scott Wiener, targeted high-cost AI models and faced opposition from Silicon Valley, including OpenAI and AI experts such as Yann LeCun. “While well-intentioned, SB 1047 does not take into account whether an AI system is deployed in high-risk environments, involves critical decision-making, or the use of sensitive data. Instead, the bill applies stringent standards to even the most basic functions — so long as a large system deploys it. I do not believe this is the best approach to protecting the public from real threats posed by the technology,” Newsom said.

California's controversial AI bill - Sync #481 Conrad Gray · Aug 23 Read full story

How AlphaChip transformed computer chip design
Google DeepMind has open-sourced AlphaChip, an AI system that uses reinforcement learning to design computer chips. Since its release in 2020, AlphaChip has been used to design superhuman chip layouts for Google’s Tensor Processing Units (TPUs), which Google employs to train and run its generative AI systems, from large language models, such as Gemini, to image and video generators, Imagen and Veo. AlphaChip generates superhuman or comparable chip layouts in hours, rather than taking weeks or months of human effort, and these layouts are already being used in data centres and mobile phones.

World’s first AI art museum to explore ‘creative potential of machines’ in LA
Refik Anadol, a prominent AI artist, is set to open Dataland, the world’s first AI art museum, in Los Angeles by late 2025. The purpose of the museum is to highlight the “intersection of human imagination and the creative potential of machines.” Anadol envisions the museum as a utopian space that reinvents traditional museums for the AI era. The museum will use ethically collected datasets and prioritise transparency around energy consumption, with support from sustainable energy sources.

Brain-like Computers Tackle the Extreme Edge
One of the biggest challenges facing AI is its high energy consumption, especially in edge devices where power constraints are critical. Australian startup BrainChip offers a solution with its Akida Pico chip, a neuromorphic chip that consumes as little as 1 milliwatt of power. Akida Pico joins other neuromorphic chips on the market, such as Innatera’s T1 and SynSense’s Xylo, which also focus on low-power, edge-based AI processing. Despite the promise of neuromorphic computing, its widespread adoption has been slow. However, BrainChip remains optimistic about its future commercial potential.

Snap is introducing an AI video-generation tool for creators
Snapchat is introducing a new AI video-generation tool that allows select creators to generate videos from text prompts, with image prompts to follow soon. The tool, available in beta on the web, is currently limited to a small group of creators, with no plans for broader public access at this time. With this move, Snapchat joins YouTube and Instagram, which have previously announced similar tools for their creators.

▶️ The dark side of AI voice cloning (13:05)

In this video, Jeff Geerling, a popular tech YouTuber, shares his experience of having his voice cloned without his consent and explains how the situation was ultimately resolved. He then discusses how easy it is to clone someone's voice and highlights the ethical issues this technology raises if misused.

Why the Training of Generative AI Models Is a Violation of Copyright Law
Tobias Mark Jensen, in his Futuristic Lawyer newsletter, analyzes a recent paper that presents a compelling argument for why AI companies should either compensate copyright holders or seek permission to use their data for training AI models. He explains that EU copyright law includes a three-stage test to determine whether exceptions to copyright law are justified, and how generative AI model training likely fails to meet these criteria. The article also stresses the importance of stricter regulations to safeguard human creativity, which may be at risk of being overshadowed or replaced by AI-generated content.

▶️ The Rise of Virtual Humans — and What They Mean for the Future (10:27)

In this TED Talk, Sara Giusto, a talent manager, discusses managing virtual influencers—digital-only social media stars. Despite being virtual, these influencers engage with their audiences and real-world issues like climate change and collaborate with major brands such as Amazon and Porsche. Sara emphasises how virtual human technology enhances creativity, connection, and expression across industries, viewing it as a revolutionary blend of technology and emotion. She urges moving away from dystopian views of AI to focus on its potential for fostering human connection and positive impact, with plans to use AI to make virtual humans more autonomous.

The rise of virtual influencers Conrad Gray · Feb 12 Read full story

If you're enjoying the insights and perspectives shared in the Humanity Redefined newsletter, why not spread the word?

Refer a friend

🤖 Robotics

Fourier launches GR-2 humanoid, software platform

Fourier revealed an updated version of their humanoid, GR-2. GR-2 is taller and heavier than GR-1, standing 175 cm (68.9 in.) tall and weighing 63 kg (139 lb.), with 53 degrees of freedom and a single-arm load capacity of 3 kg. It features a swappable battery with double the capacity and a runtime of up to two hours. Additionally, GR-2 has a new hand with 12 degrees of freedom, equipped with six tactile sensors that allow it to sense force and identify object shapes and materials. Fourier also introduced an SDK compatible with ROS and other frameworks like NVIDIA Isaac Lab and Mujoco, providing developers with pre-optimised modules for machine vision, path planning, and force feedback. There is no publicly available information about the cost of the robot.

Robot hand can detach from arm, crawl over to objects, and pick them up

Some roboticists look to nature for inspiration. Others asked, “What if we don’t?” and the result is this robot, which can detach its hand, allowing it to crawl and grasp objects independently. Also, its fingers can bend backwards.

Humanoid robotics developers must pick which problems to solve, says NVIDIA
In this interview, Deepu Talla, vice president of robotics and edge computing at NVIDIA, discusses the growing demand for humanoid robots due to labour shortages and the need for automation in dangerous jobs. He highlights two key inflection points driving this development: generative AI and advancements in digital twins and the industrial metaverse. Talla also explains NVIDIA's role in this field as a provider of tools and platforms, such as Omniverse and Project GR00T, to help developers train AI models and simulate humanoids.

Will humanoid robots take off? Conrad Gray · Jun 11 Read full story

🧬 Biotechnology

Two Nobel Prize winners want to cancel their own CRISPR patents in Europe
Emmanuelle Charpentier and Jennifer Doudna, who won the Nobel Prize for developing CRISPR, are requesting the cancellation of two of their own European patents following a damaging opinion from a European technical appeals board. The board ruled that their earliest patent filing didn’t sufficiently explain CRISPR for others to replicate, raising doubts about the validity of their invention. This marks another chapter in the decade-long CRISPR patent battle between Charpentier, Doudna, and Feng Zhang of the Broad Institute. The Broad Institute holds the dominant US patents on CRISPR, while the Nobel laureates have relied on their European patents. The cancellation could influence the complex web of biotech companies seeking “freedom to operate” in CRISPR-related research and product development and may force some of them to reassess their legal standing and coverage.

The Great Pharma Wasteland
The pharmaceutical industry is highly inefficient, with drug development costs ranging from millions to billions of dollars, taking over a decade, and with a low success rate. In this insightful article, Alex Zhavoronkov, co-founder and co-CEO of Insilico Medicine, suggests several ideas to improve the industry, including standardised metrics to evaluate the productivity and decision-making of pharma executives, similar to scientific metrics like the H-index, to enhance transparency and accountability in R&D. He highlights the frequent turnover of C-level executives and R&D leaders as a major cause of inefficiency, often leading to programme terminations for strategic reasons rather than scientific merit. Zhavoronkov also urges both pharmaceutical and AI companies to think long-term and prioritise patients over career-driven decisions and trend-following.

Crispr-Enhanced Viruses Are Being Deployed Against UTIs
As the rise in antibiotic resistance makes bacterial infections harder to treat, especially urinary tract infections (UTIs) caused by E. coli, scientists are exploring new methods for developing the next generation of antibiotics. One company, Locus Biosciences, is working on a treatment using bacteriophages (viruses that kill bacteria) enhanced with CRISPR gene-editing technology to target and kill antibiotic-resistant bacteria. In a small Phase 2 trial, 16 women with UTIs showed rapid declines in E. coli levels, with 14 considered cured by the end of the study. Locus is now launching a larger trial with 288 participants to better assess the efficacy of the phage therapy.

💡Tangents

Fly brain breakthrough 'huge leap' to unlock human mind
Scientists have created the most detailed map of a fly brain, detailing the position, shape, and connections of all 130,000 cells and 50 million connections. This breakthrough achievement offers insights into brain function, including how neurons process information, and could help in understanding the human brain. It may also lead to significant discoveries in neuroscience and pave the way for future mapping of larger brains, including humans, within the next few decades.

Thanks for reading. If you enjoyed this post, please click the ❤️ button or share it.

Share

Humanity Redefined sheds light on the bleeding edge of technology and how advancements in AI, robotics, and biotech can usher in abundance, expand humanity's horizons, and redefine what it means to be human.

A big thank you to my paid subscribers, to my Patrons: whmr, Florian, dux, Eric, Preppikoma and Andrew, and to everyone who supports my work on Ko-Fi. Thank you for the support!

My DMs are open to all subscribers. Feel free to drop me a message, share feedback, or just say "hi!"

Message Conrad Gray