What’s up with Figure? - Sync #523
Plus: OpenAI hits $10B in ARR; o3-pro is out; Disney and Universal sue Midjourney; Apple execs face questions over Siri and Apple Intelligence delays; Unitree eyes $2B pre-IPO valuation
Hello and welcome to Sync #523!
This week, we will take a closer look at Figure AI and ask: What’s going on with Figure, in light of the recent increase in scrutiny?
In other news, OpenAI hits $10 billion in ARR, releases o3-pro, delays its open-source model, and strikes a deal with Google Cloud for more computing power. Meanwhile, Midjourney is being sued by Disney and Universal; Apple executives face questions over Siri and Apple Intelligence delays; the rise of “vibe hacking”; and Meta builds a new superintelligence team, poaching Scale AI’s CEO to lead it.
Over in robotics, Unitree eyes pre-IPO fundraising at a $2 billion valuation, Aldebaran shuts down, and the UK government plans to fast-track the rollout of autonomous vehicles.
Additionally, this week’s issue of Sync features a gene-editing human embryos startup funded by the founder of Coinbase; a comparison of Culture vs the Ekumen as models for humanity’s future; why humanoid robots need their own safety rules; a breakthrough in the search for an HIV cure; and much, much more!
Enjoy!
What’s up with Figure?
In the world of technology, some companies seem to appear overnight. They capture imaginations, attract billions in investment, and redefine what’s possible. For artificial intelligence, that company was OpenAI; for humanoid robotics, it’s Figure AI. Much like OpenAI, Figure appeared seemingly out of nowhere and, in a very short time, reshaped an entire industry, quickly becoming the centre of attention in the robotics world.
Yet behind the headlines and viral demo videos, a growing chorus of sceptics is asking how much of Figure’s story is real, and what it means for the future of robotics.
Figure’s rise to a $2.6 billion valuation
Launched in 2022, Figure AI set out with an audacious goal: to build commercially viable humanoid robots that could work alongside humans. Within just a few years, the company evolved from a bold idea to a global sensation, thanks to its futuristic robots, high-profile backers, and ambitious vision for the future of work.
Figure’s robots, with their metallic grey bodies and glossy black featureless faces, are designed to stand out both visually and functionally. These machines are meant not as lab prototypes, but as practical tools for warehouses, factories, and potentially even homes.
Much of Figure’s momentum was fuelled by two high-profile partnerships. In early 2024, the company announced a collaboration with OpenAI to integrate advanced AI models into its robots, and a commercial partnership with BMW to deploy humanoids at the automaker’s South Carolina factory.
These deals caught the attention of top investors. At the same time, Figure closed a massive $675 million Series B round from the likes of Microsoft, OpenAI, Nvidia, Intel Capital, and Jeff Bezos, pushing its valuation to $2.6 billion—an extraordinary milestone for a startup just two years old.
In 2025, Figure unveiled Helix, its own Vision-Language-Action (VLA) model, marking the end of the OpenAI partnership. The company also launched BotQ, a manufacturing facility aiming to produce 12,000 robots per year per line, with ambitious plans to ship 100,000 robots in the near future. Meanwhile, rumours of a $1.5 billion Series C round at a $39.5 billion valuation have emerged, and CEO Brett Adcock revealed the signing of a second customer—possibly UPS, according to Bloomberg.
Momentum meets scepticism
For a time, Figure AI’s story was one of unstoppable momentum. The company’s futuristic robots, big-name partnerships, and record-breaking investments painted a picture of inevitable success. But as the hype around Figure grew bigger, so did the questions surrounding it.
First, Fortune took a closer look at Figure’s headline with BMW. According to the report, while Adcock publicly spoke of a “fleet” of humanoid robots carrying out “end-to-end operations” at BMW’s South Carolina factory, the reality appeared far less dramatic. BMW spokespeople told Fortune that only a single Figure robot was ever operating in the plant at any given time, and its tasks were limited, at first confined to off-hours, simply moving parts in a controlled setting. When pressed, BMW would not clarify whether there was ever more than one robot, or why the scope differed from Figure’s narrative. In response to that article, Adcock wrote on X that Figure “won’t stand for the mischaracterizations and downright lies” and announced legal actions against Fortune.
The Wall Street Journal followed up with its reporting, focusing on the reportedly ongoing $1.5 billion funding round. Figure is seeking this massive Series C investment at a $39.5 billion valuation despite, according to the WSJ, having no meaningful revenue and only a handful of robots actually in the field. The report states that investor materials leaned heavily on polished demonstration videos and ambitious forecasts—including a projection of $9 billion in revenue by 2029—but provided little in the way of audited financial statements or concrete commercial results.
Brett Adcock faces questions at Bloomberg Tech
Against a backdrop of growing scepticism, Figure AI CEO Brett Adcock made a rare public appearance at the Bloomberg Tech conference. Adcock described Figure’s robots as “synthetic humans” and declared his aim to create “the biggest product launch ever.” He positioned Figure as a company building a vertically integrated, full-stack robotic workforce, not just hardware. “We want robots to go out, do your dishes, laundry, manufacturing, construction—everything,” he said, painting a future where robots handle both industrial and domestic tasks.
Pressed on controversial topics, Adcock largely sidestepped specifics. On the much-discussed BMW partnership, he emphasised the technical learning process rather than the commercial or operational details. When asked why Figure had not brought a robot to the event, unlike competitors Agility Robotics and Boston Dynamics, who both showcased their robots live, Adcock was unapologetic, insisting that time was better spent developing the technology than putting on public shows.
When the conversation turned to Figure’s headline-grabbing funding ambitions and $39.5 billion valuation, Adcock emphasised the company’s financial strength. He stated, “We probably have on balance sheet today more cash than the collective humanoid groups have raised throughout history.” For Adcock, this capital advantage is crucial—not just for building robots at scale, but also for attracting top talent and maintaining Figure’s lead in the race to create a commercially viable humanoid workforce.
I recommend taking the time to watch the interview. The conversation touches on several interesting ideas, such as the possibility that the humanoid robotics sector might be a “winner takes all” market, and offers valuable insight into the ambitions and challenges facing Figure.
High stakes for humanoid robotics
As scrutiny mounts and doubts surface about the company’s actual achievements, the stakes extend far beyond Figure itself. Investors and rival startups alike are watching closely. If Figure delivers on its promises, the entire sector could see a surge of confidence, unlocking fresh investment and accelerating real-world adoption. However, if the company stumbles—if the hype does not match the reality—the disappointment could spread quickly, triggering a “humanoid winter.” Like the AI winters of decades past, such a downturn could see funding dry up, talent scatter, and progress slow across the board.
Last week, Figure uploaded an hour-long video to their YouTube channel of one of their robots handling various packages. This “robot ASMR” video perfectly demonstrates the Figure robot’s near-human-level dexterity and speed when sorting parcels. It is remarkable that machines like this now exist (that box flip at 1:05 is particularly impressive). But as impressive as these demonstrations are, the future of Figure—and perhaps humanoid robotics as a whole—now depends on turning bold promises into reality. Whether Figure’s story will become the blueprint for a new era or a cautionary tale remains to be seen, but one thing is certain: the world is watching, and the next steps will shape the path for everyone in the field.
If you enjoy this post, please click the ❤️ button or share it.
Do you like my work? Consider becoming a paying subscriber to support it
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.
🦾 More than a human
Crypto billionaire Brian Armstrong is ready to invest in CRISPR baby tech
Coinbase CEO Brian Armstrong has announced plans to fund a US startup dedicated to gene-editing human embryos, marking the first major commercial move into a controversial area of medicine previously stigmatised and underfunded. Armstrong is seeking scientists to build a company focused on correcting genetic diseases using new, safer base-editing technology, but his ambition comes amid strict legal bans and fierce ethical debate over “designer babies” and heritable genome editing. Despite the potential for medical breakthroughs, the technology remains highly controversial, with leading organisations advocating for a decade-long ban to allow for further study and debate.
▶️ How artificial retinas could cure blindness (11:02)
In his TEDx talk, Dante Muratore, a professor in bioelectronics at Delft University of Technology, explores the transformative power of brain-computer interfaces (BCIs), focusing on an artificial retina his team is developing to restore sight in patients with degenerative eye diseases like macular degeneration and retinitis pigmentosa. He explains both the technical and ethical challenges of creating a device that can replicate the brain’s natural neural code, discusses the wider potential of BCIs to treat other neurological disorders, and urges society to reflect on how this technology should be used for the benefit of all.
🔮 Future visions
▶️ The Culture vs The Ekumen as Models for Humanity's Future (19:16)
Hank Green reflects on the contrasting visions of the future found in Iain M. Banks’ “Culture” novels and Ursula K. Le Guin’s “Ecumen” stories. He offers a thoughtful meditation on how both authors imagine humanity’s future—Banks envisions a highly advanced, interventionist, post-scarcity society seeking to uplift others, while Le Guin depicts a loosely connected, diverse confederation that values observation, diversity, and non-intervention. Hank finds both approaches compelling but contradictory, pondering whether humanity is heading towards a unified, technologically driven future or a web of coexisting differences, and connects these ideas to contemporary debates about technological change and cultural diversity.
🧠 Artificial Intelligence
OpenAI hits $10 billion in annual recurring revenue fueled by ChatGPT growth
OpenAI has reached $10 billion in annual recurring revenue in less than three years after launching ChatGPT, according to a company spokesperson. The figure excludes revenue from Microsoft licensing and large one-off deals. Despite losing $5 billion last year, the company is targeting $125 billion in revenue by 2029, according to a report by The Information.
OpenAI taps Google in unprecedented cloud deal despite AI rivalry
Reuters reports that OpenAI has struck a deal to use Google Cloud services to meet surging demand for computing power. Finalised in May, the agreement helps OpenAI diversify beyond its main backer, Microsoft, and is seen as a major win for Google's cloud division. The partnership also involves CoreWeave, a neocloud company backed by Nvidia, which will provide GPU-based computing capacity to Google Cloud. Google will then resell this capacity to OpenAI, reflecting the growing web of alliances forming across the AI industry to meet escalating infrastructure needs.
Disney and Universal Sue AI Company Midjourney for Copyright Infringement
Disney and Universal have filed a landmark copyright lawsuit against Midjourney, a popular AI image generator, accusing it of mass-producing unauthorised copies of their intellectual property. The studios allege Midjourney ignored requests to implement safeguards and illegally copied protected content during training. This marks the first time major Hollywood players have taken direct legal action against an AI company, potentially setting a pivotal precedent in the ongoing debate over copyright and AI-generated content.
OpenAI launches o3-pro
OpenAI has launched o3-pro, a model that OpenAI calls its most intelligent yet. o3-pro is now available to ChatGPT Pro and Team users, replacing o1-pro. Enterprise and Edu users will gain access next week (note the absence of the Plus plan mentioned). It is also available via OpenAI’s API at a rate of $20 per million input tokens and $80 per million output tokens. According to OpenAI, the new model offers enhanced performance across clarity, accuracy, and instruction-following. Image generation, temporary chats and Canvas are currently not available with o3-pro.
▶️ Apple Execs Defend Siri Delays, AI Plan and Apple Intelligence (24:13)
It’s fair to say that the introduction of Apple Intelligence did not go as planned. In this interview, Craig Federighi and Greg Joswiak faced tough questions about the delays to the new Siri and what’s next for Apple Intelligence—questions many Apple users would no doubt like to ask the company’s leadership. Federighi and Joswiak admitted that reliability issues prevented the technology from meeting Apple’s high standards and confirmed it was functional but not ready for broad release. The executives defended the delay as a necessary move to maintain quality, insisting that Apple’s long-term AI strategy prioritises thoughtful innovation over being first to market.
Updates to Apple's On-Device and Server Foundation Language Models
At WWDC 2025, Apple announced a new Foundation Models framework, which will open its on-device foundation language model to app developers, enabling them to power their own AI features with improved performance, tool use, efficiency, and support for 15 languages, according to Apple. Alongside this announcement, Apple also published an update on its foundation models, including benchmark results comparing its on-device and server models to competitors such as Qwen 3 and Gemma 3 (for the on-device model), and Llama 4 Scout and GPT-4o (for the server model). Despite technical advances and a strong emphasis on privacy and responsible AI, Apple’s models were often rated as only “comparable” to or below those of rivals, with human testers preferring outputs from OpenAI’s GPT-4o and Meta’s Llama 4 Scout in several key tasks.
Zuckerberg Is Personally Recruiting New ‘Superintelligence’ AI Team at Meta
Frustrated by Meta’s lagging performance in AI, Mark Zuckerberg is personally assembling a high-level “superintelligence” team to pursue artificial general intelligence (AGI), with the aim of surpassing competitors such as OpenAI and Google. According to Bloomberg, Zuckerberg is recruiting around 50 experts, including a new head of AI research. The plan also includes a $14.3 billion investment in Scale AI and the hiring of its co-founder, Alexandr Wang, to help lead Meta’s new AGI lab. Zuckerberg’s hands-on approach follows internal disappointment with Meta’s Llama 4 model and delays in the rollout of its next AI system, Behemoth (more on that here), as he pushes to position Meta as a global leader in AI.
OpenAI’s open model is delayed
OpenAI has delayed the release of its anticipated open-weight AI model until later this summer. Sam Altman wrote in a post on X that the OpenAI team “did something unexpected and quite amazing” and that the delay is worthwhile.
▶️ Can AMD match NVIDIA in 2025 or 2026? (12:32)
In this video, Dr Ian Cutress breaks down AMD’s latest AI hardware roadmap, starting with the MI355X, which aims to undercut Nvidia in AI cost-efficiency while doubling performance over its predecessor. He also previews the upcoming MI400, expected in 2026, which promises a 4x performance boost, HBM4 memory integration, and Ultra Accelerator Link technology to enable scaling across up to 1,024 GPUs. Alongside these advancements, AMD is rolling out ROCm 7—a major software update offering day-one support for key AI frameworks—and introducing a full-stack rack-scale solution including CPUs, GPUs, and AI networking. With these developments, AMD is directly challenging Nvidia’s dominance in the hyperscale AI market.
Magistral—the first reasoning model from Mistral AI
Mistral, a French AI lab, has released its first reasoning model, named Magistral. The new model comes in two variants—Magistral Small, a 24B parameter open-source version, and Magistral Medium, a closed enterprise version. According to Mistral, the new models offer multilingual capabilities, step-by-step logic, and real-time responses in Le Chat (Mistral’s equivalent to ChatGPT), making it ideal for high-stakes sectors like law, finance, and engineering. Magistral Small is available on Hugging Face. Magistral Medium is available on Mistral’s platform and via various cloud providers.
Introducing the V-JEPA 2 world model and new benchmarks for physical reasoning
Meta has introduced V-JEPA 2, an advanced AI "world model" designed to help robots and AI agents understand and interact with the physical world more intuitively. Building on last year’s V-JEPA, which was trained on over a million hours of video, the new model enables predictions of real-world outcomes, like anticipating the trajectory of a ball or handling objects in a kitchen. Meta claims V-JEPA 2 is 30 times faster than Nvidia’s similar Cosmos model, though performance comparisons may differ by benchmarks. V-JEPA 2 is available on GitHub and Hugging Face.
The Gentle Singularity
Sam Altman believes humanity has already crossed a crucial threshold in artificial intelligence, with digital superintelligence now firmly within reach. In this post, he argues that while the world outwardly remains much the same, recent AI breakthroughs have made tools that surpass human capabilities in some respects, greatly amplifying productivity and scientific progress. Altman foresees the 2030s as a time of abundance in intelligence and energy, with new societal challenges and opportunities emerging, including the need for robust AI alignment and equitable access. He concludes that “intelligence too cheap to meter” is approaching fast, urging society to adapt quickly and collectively shape the future of superintelligence.
The Rise of ‘Vibe Hacking’ Is the Next AI Nightmare
You heard about vibe coding, but have you heard about vibe hacking? As generative AI tools become more accessible, cybersecurity experts are warning of a future where even unskilled users can launch sophisticated cyberattacks with AI’s help. Systems like XBOW, an autonomous AI hacker currently topping leaderboards on HackerOne, demonstrate how AI can already find and exploit web vulnerabilities at scale. While fully autonomous AI-led attacks aren't yet a reality, experts say the technology is rapidly evolving, arming both seasoned hackers and “script kiddies” with tools to create and deploy malicious code faster than ever.
Anthropic’s AI-generated blog dies an early death
Anthropic has shut down Claude Explains, a blog showcasing Claude’s writing capabilities, just weeks after launching it. The blog, which featured human-edited posts on technical topics, aimed to blend customer education with marketing but faced criticism over its lack of transparency about AI-generated content.
AI to play increasing role in UK armed forces, defence secretary says
The British Defence Secretary, John Healey, has announced that AI will be central to Britain’s upcoming strategic defence review, which aims to modernise the armed forces and avoid past procurement delays. The review, developed over the past year, will focus on rapidly deploying advanced technologies such as AI and drones, with concrete procurement plans to be unveiled within weeks. Amid rising global threats and lessons from Ukraine, Healey stressed the need to equip frontline troops faster and innovate at a “wartime pace”. The government also plans to allocate 10% of the military equipment budget to emerging tech and potentially boost partnerships with firms like Palantir.
🤖 Robotics
China’s Unitree eyes pre-IPO fundraising, valuing robot maker up to US$2 billion
Unitree Robotics, a Chinese robotics company known for its G1 humanoid robots and a series of robot-dogs, is reportedly preparing a new fundraising round that could value the company at up to 15 billion yuan (US$2.1 billion), fuelling speculation about an impending IPO. Unitree has gained visibility through public performances and is among several Chinese robot makers pushing for the mass production of humanoid robots in 2025.
Aldebaran, maker of Pepper and Nao robots, put in receivership
Aldebaran, a French robotics company known for its Nao and Pepper humanoid robots, has filed for bankruptcy and entered receivership following years of financial difficulties, layoffs, and failed acquisition attempts. Founded in 2005 and acquired by SoftBank in 2012, Aldebaran was once a leader in educational and service robotics but struggled with declining demand and outdated technology. After its current owner, United Robotics Group, ceased funding in 2024, the company began seeking a buyer, leaving the future of its staff and intellectual property uncertain. According to The Robot Report, the company is still looking for a buyer. However, a farewell video uploaded to Aldebaran’s YouTube channel states that the company is shutting down and thanks everyone who has been part of its 20-year journey.
Why humanoid robots need their own safety rules
As humanoid robots begin operating in warehouses, experts are pushing for new safety standards to address the unique physical and psychological risks they pose. Unlike traditional robots, humanoids require constant power to stay upright, making emergency stops potentially dangerous. The IEEE Humanoid Study Group and ISO are developing guidelines to ensure these machines can safely work alongside humans, emphasising gradual responses to danger, clearer communication signals, and realistic expectations of robot abilities.
The UK Accelerates Its Self-Driving Car Ambitions
The UK government has announced it will fast-track the rollout of autonomous vehicles, allowing limited pilot programmes on public roads starting in spring 2026—a year earlier than previously planned. Citing economic growth, job creation, and improved transport and safety, the UK Transport Secretary said the industry could contribute £42 billion and 38,000 jobs by 2035. Companies like Wayve and Uber are already preparing for trials in London, with full commercial launches expected in 2027. Safety will remain a top priority, with strict testing required before broader deployment.
Walmart and Wing expand drone delivery to five more US cities
Wing, the Alphabet-owned drone delivery company, is expanding its partnership with Walmart, announcing plans to offer drone deliveries from over 100 Walmart stores across five new US cities—Atlanta, Charlotte, Houston, Orlando, and Tampa—alongside its existing operations in the Dallas-Fort Worth area. This move marks Wing's shift from pilot programmes to broader implementation and commercial scaling.
From festivals to weddings: Why drone shows are booming
Drone shows are rapidly gaining popularity as a high-tech alternative to fireworks, offering creative and emotionally resonant visual spectacles at events like weddings, festivals, and major public celebrations. While these shows offer precision, synchronisation with music, and custom storytelling potential, their high cost and logistical complexity remain challenges. Nonetheless, the growing sophistication and impact of drone displays suggest a bright future for this emerging art form.
🧬 Biotechnology
Breakthrough in search for HIV cure leaves researchers ‘overwhelmed’
Researchers from Australia have made a major breakthrough in the search for an HIV cure by demonstrating, for the first time, that mRNA can be delivered into the white blood cells where HIV hides, forcing the virus to be exposed—something previously thought impossible. The team then developed a new type of lipid nanoparticle capable of targeting these cells, a key step toward flushing out the virus, though further animal and human trials will be needed.
Anne Wojcicki to buy back 23andMe and its data for $305 million
Anne Wojcicki, co-founder of 23andMe, is set to reacquire most of the company’s assets for $305 million through her nonprofit, TTAM Research Institute, outbidding Regeneron’s previous offer of $256 million. The deal, still pending court approval, follows 23andMe’s March bankruptcy filing and includes the firm’s DNA testing and research services, as well as its telehealth unit Lemonaid Health.
Artificial Blood That Could Work for All Blood Types in Trials
Researchers in Japan have begun clinical trials of universal artificial blood, which, if successful, could help address global blood shortages. This haemoglobin-based product, compatible with all blood types and storable for up to two years, was transfused into 16 healthy volunteers in March. If proven safe and effective, it could transform emergency care and transfusion medicine, particularly in low-income countries with limited access to donor blood. Broader trials are planned, with hopes for clinical use by 2030.
Thanks for reading. If you enjoyed this post, please click the ❤️ button or share it.
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!"