On June 29, 2025, Daniel Gross walked away from a company valued at $32 billion—one he had co-founded just thirteen months earlier with Ilya Sutskever, the legendary OpenAI co-founder and chief scientist. The destination: Meta Superintelligence Labs, Mark Zuckerberg's newly assembled dream team dedicated to building artificial general intelligence.
The move completed a remarkable journey for the 33-year-old Israeli-American. In the span of fifteen years, Gross had evolved from a self-taught programmer in Jerusalem's Orthodox community to one of the most influential figures in artificial intelligence—not through building AI systems himself, but by identifying, funding, and nurturing the people and companies that did.
"I expect miracles to follow," Gross wrote in his farewell to Safe Superintelligence Inc. The statement captured both his characteristic optimism and the audacious scale of his ambitions.
The timing of Gross's departure illuminated a pivotal moment in the AI industry. When even quadrupling a $1.1 billion venture fund in two years—as Gross and his partner Nat Friedman had done with NFDG—seemed less compelling than direct involvement in AI development, something fundamental had shifted in Silicon Valley's calculus. The Age of AI demanded operators, not just investors.
For Gross, the transition represented a return to his roots. Before he became one of tech's most prolific angel investors, before he built Y Combinator's AI program, before he deployed $100 million in GPU clusters to nurture AI startups, he had been a builder—a teenager who created search engines in Jerusalem's military academies and sold a company to Apple at 21.
This is the story of how an Orthodox Jewish teenager from the Katamon neighborhood of Jerusalem became the architect of Silicon Valley's AI future—and what his trajectory reveals about the rapidly evolving landscape of artificial intelligence.
Daniel Gross was born in Jerusalem in 1991 to American parents who had emigrated to Israel. He grew up in Katamon, a neighborhood in southwestern Jerusalem known for its mix of secular and religious residents. Raised in an Orthodox Jewish household, Gross expected to lead a traditional religious life somewhere in Israel.
"I spent most of my youth feeling like an outsider looking in," Gross later wrote. "High school wasn't interesting. I didn't have many friends."
The outsider mentality would prove formative. While his peers followed conventional paths through Israeli religious education and military service, Gross found escape in computers. At 13, he began teaching himself to program. By 18, he had built several projects that demonstrated unusual aptitude—all without formal computer science training.
Gross attended the Horev yeshiva in Jerusalem, a religious school that combined Talmudic study with secular education. He then enrolled in the Eli pre-military academy, a program designed to prepare young Israelis for leadership positions in the Israel Defense Forces. The trajectory seemed clear: complete the pre-army program, serve in the IDF, and perhaps attend university afterward.
But during his time at the Eli academy, Gross discovered Y Combinator. The legendary startup accelerator, founded by Paul Graham in 2005, had begun accepting applications from international founders. On a whim, Gross applied.
The circumstances of Gross's Y Combinator application have become Silicon Valley legend. As he later recounted, he connected his "trusty Nokia cell phone to a clunky laptop and applied to YC from the desolate Israeli military camp where I was based."
The startup he pitched was Greplin, a personal search engine that would allow users to search across their online accounts—Facebook, LinkedIn, Twitter, email, and cloud storage—from a single interface. The concept anticipated the fragmentation of digital identity across platforms that would only intensify over the following decade.
Y Combinator accepted Gross into its Winter 2010 batch. At 17, he became the youngest founder ever admitted to the program.
The acceptance created an impossible choice. Israeli law requires all citizens to complete mandatory military service, typically beginning at age 18. Gross was scheduled to enlist in the IDF. But Y Combinator operated in Silicon Valley, 7,000 miles away, on an incompatible timeline.
Gross chose Silicon Valley. He never completed his IDF service.
The decision carried lasting consequences. For years, Gross could not return to Israel without facing arrest by military police. He had traded his homeland—and the life his community expected of him—for a chance to build something in California.
Greplin's concept resonated in Silicon Valley. The startup addressed a genuine problem: as users' digital lives fragmented across dozens of services, finding specific information—a photo from Facebook, an email attachment, a file in Dropbox—required logging into each platform separately.
In 2011, when Gross was 19, Sequoia Capital led a $4 million Series A round in Greplin. The investment marked a watershed moment. Sequoia, which had backed Apple, Google, YouTube, Instagram, and WhatsApp, rarely bet on teenagers. Gross became one of the youngest founders in Sequoia's portfolio history.
The vote of confidence from Silicon Valley's most prestigious venture firm validated not just Greplin, but Gross's unconventional path. He had left religious education in Jerusalem, skipped military service, and built something that Sequoia considered worth millions.
In 2012, Greplin rebranded as Cue and expanded its vision. Beyond search, the company would predict what users needed—surfacing relevant information proactively based on context. Cue raised an additional $10 million from Index Ventures in November 2012.
The product evolved into a "personal assistant" that pulled information from users' online accounts to present an overview of their day. Calendar appointments, email summaries, social updates, and travel information appeared in a unified interface. The vision was prescient: it anticipated the AI-powered assistants that would emerge years later.
In October 2013, Apple acquired Cue for an undisclosed amount estimated between $40 million and $60 million. Gross was 21 years old.
The acquisition reflected Apple's growing anxiety about artificial intelligence. Google had launched Google Now in 2012, an AI-powered assistant that could predict users' needs based on their data. Amazon had introduced Alexa development internally. Apple's Siri, acquired in 2010 and launched in 2011, was falling behind.
Apple saw in Cue the technology it needed to enhance Siri's capabilities. The contextual search technology—which could pull relevant information from multiple sources and predict user intent—aligned precisely with Apple's vision for intelligent assistance.
More importantly, Apple saw in Gross a technical leader who understood both search technology and machine learning. The company shut down Cue immediately after the acquisition, but retained the entire team.
Gross joined Apple as a director, overseeing AI and search projects across iOS, macOS, and watchOS. At 22, he was leading teams responsible for some of Apple's most technically challenging work.
The role offered an extraordinary vantage point. Apple was in the early stages of integrating machine learning throughout its products—not just in Siri, but in photo recognition, keyboard predictions, battery optimization, and dozens of other features. Gross witnessed firsthand how a company with billions of devices could deploy AI at scale.
During his four years at Apple (2013-2017), Gross worked on projects that would define the company's AI strategy. He helped develop features that intelligently pulled contact information from correspondence and integrated it across apps. He contributed to Spotlight's evolution from a simple search tool into an intelligent assistant capable of understanding context.
The experience shaped Gross's understanding of AI's commercial potential. At Apple, he saw that machine learning wasn't just an academic curiosity—it was becoming the substrate of modern software. Every product, every feature, every user interaction could be enhanced by intelligent systems.
Yet Apple also revealed the constraints of building AI within a large corporation. The company's famous secrecy, while protecting product development, limited collaboration with the broader AI research community. Apple's privacy-first philosophy, while admirable, restricted the data available for training machine learning models.
Most fundamentally, Apple moved at Apple's pace. A startup could pivot overnight; Apple required years of planning, review, and coordination across hundreds of teams. For someone who had built and sold a company by 21, the deliberate cadence of big tech felt constraining.
By 2016, Gross was contemplating his next move. He had proven he could build within a startup and execute within a giant corporation. But neither model felt optimal for the AI revolution he saw coming.
In late 2016, Gross received an unexpected invitation. Y Combinator—the accelerator that had accepted his application from an Israeli military camp seven years earlier—wanted him to return as a partner.
In January 2017, Y Combinator announced that Daniel Gross would join as a partner, leaving Apple after four years. The news generated unusual attention in tech circles. Gross wasn't just another operator joining a VC firm; he represented a direct connection between Silicon Valley's AI research community and its startup ecosystem.
"Daniel led search and AI at Apple after Apple acquired his startup Cue," Y Combinator's announcement noted. "Before that, he was the youngest entrepreneur we ever funded."
The return completed a narrative arc. The 17-year-old who had applied from an Israeli military camp was now a partner at the organization that had launched his career. But Gross had no interest in simply reviewing applications and mentoring founders. He came with a specific mission: make Y Combinator the definitive launchpad for AI companies.
In March 2017, two months after joining, Gross announced Y Combinator's first "vertical" track: YC AI. The program would provide specialized support for artificial intelligence startups beyond YC's standard offering.
The vertical included dedicated resources:
The initiative reflected Gross's diagnosis of AI startups' unique challenges. Unlike software companies that could launch with minimal infrastructure, AI startups required massive compute resources to train models. They needed technical expertise that most accelerators couldn't provide. And they faced competitive pressure from tech giants that were hoarding AI talent and data.
"We want to democratize AI," Gross explained. "We want to level the playing field for startups to ensure that innovation doesn't get locked up in large companies like Google or Facebook."
Gross's approach to YC AI revealed sophisticated thinking about where AI could create value. He prioritized three categories:
Perception: Companies using AI to understand the physical world. This included Standard Cognition (automating store checkout), VergeSense (facility management), CureSkin (classifying skin conditions), Modular Science (robotic farming), and D-ID (obfuscating faces for security).
Autonomy: Companies building systems that could act independently in the physical world—self-driving vehicles, robots, drones.
ML Services: Companies providing infrastructure and tools for other businesses to deploy machine learning.
The framework anticipated the AI industry's evolution. Perception companies would benefit from improvements in computer vision. Autonomy companies would capitalize on robotics advances. ML services companies would profit from every other business adopting AI.
Gross spent eighteen months at Y Combinator, mentoring companies in the Winter 2017 and subsequent 2018 batches. He provided strategic guidance on product development, scaling challenges, and the pitfalls of AI entrepreneurship.
But by mid-2018, Gross was questioning the accelerator model itself. YC was concentrated in Silicon Valley, limiting its reach. The program's batch structure—intensive periods followed by demo days—didn't align with how many successful companies developed. And the standard terms, while founder-friendly, didn't allow for the deep, long-term relationships Gross believed best companies required.
What if the accelerator could be reimagined for a remote-first world? What if it could find talent anywhere on earth, not just among those who could afford to move to San Francisco?
In August 2018, Gross left Y Combinator to find out.
Pioneer launched in August 2018 with backing from two of Silicon Valley's most influential figures: Marc Andreessen and Patrick Collison's Stripe. The concept was simple but radical: identify ambitious people anywhere in the world and give them the resources to build.
"In the way software is eating the world, remote is almost eating earth in the sense that it may very well be the way large companies are created, but also perhaps the way that venture funding takes place," Gross explained.
The thesis resonated with Gross's personal history. He had been an outsider in Jerusalem, far from Silicon Valley's resources and networks. Only the internet—and Y Combinator's willingness to accept international applications—had enabled his career. How many other talented people were isolated in places without startup ecosystems?
Pioneer operated differently from traditional accelerators. Rather than accepting cohorts for intensive programs, it ran continuous tournaments where applicants competed on projects. Winners received funding, mentorship, and access to Pioneer's network. The entire process happened online.
Pioneer's economics reflected its experimental nature. While Y Combinator invested $150,000 for 7% equity, Pioneer typically invested around $20,000 for 5% equity plus an additional 1% for program participation.
The smaller investment size allowed Pioneer to make far more bets. Over its lifetime, the accelerator funded over 150 companies with more than 300 founders in 50+ countries. The portfolio eventually exceeded $1 billion in aggregate value, with companies raising over $200 million from firms including Sequoia, Andreessen Horowitz, General Catalyst, and Y Combinator itself.
Pioneer also served as a laboratory for ideas that would inform Gross's later ventures. The continuous assessment model—rather than point-in-time evaluations—presaged the ongoing relationships he would build with AI founders through NFDG. The global reach anticipated the talent networks he would leverage for Safe Superintelligence.
While building Pioneer, Gross continued developing his personal investment portfolio. He had begun angel investing in 2011, shortly after Greplin's Sequoia funding, and never stopped.
His track record became remarkable. Gross invested early in:
By 2020, Gross had completed over 90 angel investments, an extraordinary volume for an individual investor. The portfolio's success rate—with multiple unicorns and several acquisitions—suggested either exceptional judgment or exceptional access to deal flow. In reality, it was both.
Nat Friedman's path to Silicon Valley paralleled Gross's in some ways and diverged in others. Born in 1977, Friedman was fourteen years older than Gross. He had co-founded Ximian, a Linux software company, in 1999 and sold it to Novell in 2003. He later co-founded Xamarin, a cross-platform mobile development tool, which Microsoft acquired in 2016 for approximately $500 million.
After the Xamarin acquisition, Friedman joined Microsoft, eventually becoming CEO of GitHub after Microsoft acquired the developer platform for $7.5 billion in 2018. He served as GitHub's CEO until November 2021.
Gross and Friedman had been acquainted for years—Gross had invested in GitHub before its acquisition—but their relationship intensified around 2021, as both became convinced that artificial intelligence represented the most important technological shift since the internet.
The partnership combined complementary strengths. Friedman brought deep operational experience from leading GitHub's 2,500-employee organization. Gross brought his network of AI founders and his understanding of the startup landscape. Both shared a belief that AI startups faced unique challenges requiring unique solutions.
Their first major joint initiative was AI Grant, a non-profit program providing funding to AI researchers and startups. Established in 2017 (though it scaled significantly after Friedman's involvement), AI Grant offered $250,000 in funding plus $250,000 in Microsoft Azure cloud credits to selected companies.
The program's structure reflected hard-won lessons about AI company formation. The funding came through a no-cap, no-discount MFN SAFE—maximally founder-friendly terms. Recipients gained access to a summit in San Francisco featuring advisors and founders, plus an invite-only Demo Day for showcasing to investors.
By 2024, AI Grant had incubated approximately 60 companies. The portfolio included early investments in companies that would become significant players in the AI ecosystem.
But AI Grant revealed a deeper problem. Funding alone wasn't sufficient for AI startups. They needed compute—vast quantities of GPU power to train their models. And by 2022, GPUs had become nearly impossible to obtain.
The shortage stemmed from multiple factors. NVIDIA dominated the AI chip market, and demand for its H100 GPUs far exceeded supply. Cloud providers like AWS and Google Cloud had waiting lists stretching months. Large tech companies were hoarding chips for internal projects. AI startups, even well-funded ones, couldn't access the hardware they needed.
Friedman, who had spent years at Microsoft and GitHub understanding enterprise infrastructure, and Gross, who had watched AI startups struggle to train models, conceived a radical solution: they would build their own supercomputer and offer access to their portfolio companies.
In 2023, they deployed the Andromeda Cluster. The initial configuration included 2,512 NVIDIA H100 GPUs—at a time when individual H100s were selling for $30,000-$40,000 when available at all. Including electricity, cooling, and infrastructure, the project cost approximately $100 million.
The Andromeda Cluster wasn't just a pile of GPUs. It was a sophisticated distributed computing system designed for AI workloads:
By 2024, the cluster had expanded to over 4,000 GPUs, including H100s, H200s, and B200s. The system offered multiple orchestration options—Slurm, Kubernetes, or direct SSH access to nodes.
Gross and Friedman offered access to their portfolio companies at below-market rates, effectively subsidizing the compute that AI startups desperately needed. As Friedman told Forbes, he had become "a full-time computer chip broker for upstart AI companies." During some weeks, he spent most of his time finding GPUs for people.
The compute-for-equity model created powerful alignment. Startups received resources they couldn't obtain elsewhere. Gross and Friedman gained equity in promising AI companies. The arrangement anticipated a fundamental truth about AI development: in an industry where compute was the primary constraint, those who controlled chips controlled the future.
In 2023, Gross and Friedman formalized their partnership by launching NFDG (named for their initials: Nat Friedman, Daniel Gross), a venture capital fund with an extraordinary $1.1 billion in committed capital.
The fund's size and strategy reflected the partners' conviction about AI's trajectory. NFDG led rounds from seed to growth, investing between $1 million and $100 million per deal. The focus was explicitly on AI: AI-enabled products, AI infrastructure, AI applications.
The portfolio rapidly included some of the AI industry's most valuable companies:
The returns were extraordinary. With only approximately 50% of the fund deployed, NFDG achieved roughly 4x returns—from $550 million deployed to approximately $2.2 billion in portfolio value. In just two years, the fund had more than quadrupled its investors' money.
In early 2024, Daniel Gross received a call that would redirect his career. On the other end was Ilya Sutskever, the co-founder and former chief scientist of OpenAI, who had just departed the company he helped create.
Sutskever's departure from OpenAI followed months of turmoil. In November 2023, he had been part of the board that briefly fired CEO Sam Altman, then reversed course within days. By January 2024, Sutskever had moved from the board to an advisory role. By June, he was ready to start something new.
His vision was ambitious to the point of audacity: build safe superintelligence. Not another large language model company. Not another AI application business. A company focused exclusively on developing artificial general intelligence that would be fundamentally safe for humanity.
Sutskever wanted Gross as co-founder and CEO.
The choice seemed counterintuitive. Sutskever was among the world's foremost AI researchers, a pioneer of deep learning who had worked alongside Geoffrey Hinton and Alex Krizhevsky on the breakthrough AlexNet paper. Gross had never published AI research. His expertise was in business, investing, and company building.
But that was precisely the point. Sutskever wanted to focus on the technical problems of superintelligence and AI alignment. He needed a co-founder who could handle everything else: fundraising, recruiting, operations, strategy. Gross's track record—building and selling Cue, running teams at Apple, creating Pioneer, deploying the Andromeda Cluster—demonstrated exactly those capabilities.
The third co-founder was Daniel Levy, who had led the "Optimization Team" at OpenAI. Together, the three Daniels (Sutskever's first name is Ilya, but his Hebrew name is Daniel) represented a unique combination: world-class AI research, operational excellence, and deep technical implementation experience.
Safe Superintelligence Inc. (SSI) announced its formation in June 2024. The company's mission statement was uncompromising: "Our singular focus means no distraction by management overhead or product cycles, and our business model means safety, security, and progress are all insulated from short-term commercial pressures."
The funding matched the ambition. SSI raised $1 billion in Series A financing at a $5 billion valuation—one of the largest initial fundraises in AI history. Investors included Andreessen Horowitz, Sequoia Capital, DST Global, SV Angel, and NFDG (Gross's own fund).
The company's approach differed fundamentally from other AI labs. OpenAI, Anthropic, and Google DeepMind all operated commercial businesses alongside their research efforts. They shipped products, generated revenue, and faced pressure to release capabilities quickly. SSI would do none of that.
"We approach safety and capabilities in tandem, as technical problems to be solved through revolutionary engineering and scientific breakthroughs," the company stated. "We plan to advance capabilities as fast as possible while making sure our safety always remains ahead."
By April 2025, SSI had raised an additional $2 billion at a $32 billion valuation. The round was led by Greenoaks with a $500 million commitment. Lightspeed Venture Partners and Andreessen Horowitz participated. Google's parent company Alphabet and NVIDIA had also invested, with Google Cloud becoming a major infrastructure provider.
The valuation was remarkable for a company with no products, no revenue, and approximately 20 employees. It reflected investor conviction about two things: Ilya Sutskever's unique capabilities in AI research, and the belief that whoever achieved superintelligence first would capture essentially unlimited value.
As CEO, Gross managed the company's operations while Sutskever focused on research. The division of labor mirrored their collaboration's founding premise: technical genius paired with business excellence.
In the first half of 2025, Meta Platforms approached Safe Superintelligence with an acquisition offer. The details weren't disclosed, but given SSI's valuation trajectory, the price would have been extraordinary—potentially the largest AI acquisition ever.
Sutskever refused. The company's mission—building safe superintelligence—required independence from big tech's commercial pressures. Selling to Meta would undermine everything SSI represented.
But Zuckerberg wasn't finished. If he couldn't buy the company, he would recruit its leadership.
By mid-2025, Mark Zuckerberg had declared artificial intelligence Meta's top priority. The company had already spent tens of billions of dollars on AI infrastructure—data centers, chips, power generation. But infrastructure alone wasn't sufficient. Meta needed people.
The AI talent war had reached unprecedented intensity. OpenAI, Anthropic, Google, and Meta competed for a limited pool of researchers and engineers. Compensation packages escalated into the hundreds of millions of dollars. Ruoming Pang, the engineer leading Apple's foundation models team, reportedly received $200 million over four years to join Meta.
In June 2025, Zuckerberg made his most audacious move: acquiring 49% of Scale AI and bringing founder Alexandr Wang into Meta as chief AI officer to lead a new Meta Superintelligence Labs. The deal reportedly valued Scale AI at approximately $28 billion, with Meta investing $14.3 billion.
Wang's appointment set up Meta's play for Gross and Friedman. If Meta couldn't acquire Safe Superintelligence, it could at least recruit its CEO.
The terms of Gross and Friedman's move to Meta revealed sophisticated financial engineering. Meta agreed to acquire a substantial portion of NFDG's holdings—potentially more than $1 billion worth—providing liquidity to the fund's limited partners without giving Meta control over the investments or information about the portfolio companies.
For NFDG's investors, the deal was exceptional. They had committed capital to a venture fund expecting returns over a decade. Instead, they received liquidity within two years, at a 4x multiple. Few venture investments achieve such returns; fewer still provide them so quickly.
For Gross and Friedman, the arrangement enabled their transition from investors to operators. They would join Meta Superintelligence Labs, working under Wang on the company's most ambitious AI projects. Friedman became vice president of product and applied research. Gross joined without a formal title announcement, suggesting a senior technical or strategic role.
Gross's departure from Safe Superintelligence raised questions. He had co-founded the company just thirteen months earlier. It was valued at $32 billion. Why leave?
The answer likely involves multiple factors. SSI, by design, was a research lab focused on long-term superintelligence development. It had no products, no commercial operations—just fundamental research. For someone with Gross's operational background, the role may have felt limiting.
Meta offered something different: the chance to build AI products at scale, with essentially unlimited resources. Meta's infrastructure spending dwarfed what any startup could deploy. Its distribution—billions of users across Facebook, Instagram, WhatsApp, and the metaverse—provided an unmatched platform for AI applications.
There was also the team. Working alongside Alexandr Wang, who had built Scale AI into an AI industry backbone, and Nat Friedman, his longtime partner, offered a collaboration opportunity unavailable elsewhere. The Meta Superintelligence Labs assembled some of AI's most accomplished operators and researchers under one roof.
Finally, Gross may have recognized a historical inflection. The AI industry had reached a point where building mattered more than investing. Even quadrupling a billion-dollar fund seemed less compelling than directly shaping how superintelligence developed. "I expect miracles to follow," he wrote about SSI's future under Sutskever's sole leadership. But he wanted to create miracles himself.
Daniel Gross's investment track record—over 90 investments, multiple unicorns, returns measured in billions—invites analysis. What did he see that others missed?
His explanations emphasize pattern recognition across domains. When evaluating Uber in its early days, Gross didn't just see a taxi replacement. He saw the unbundling of car ownership, the application of software to physical logistics, the creation of a new labor market. When investing in Figma, he recognized that design tools would follow the same cloud-native trajectory that other software categories had traversed.
"The equivalent of looking at the iPhone and dreaming of Uber may be very hard to predict," Gross observed. Three years after the iPhone's launch, the top apps were Facebook and games. "Everyone thought that this was what the iPhone was for; sort of a gaming product with your friends. The ideas of Uber and Instacart had not fully come around."
The observation reveals Gross's meta-level thinking. He doesn't just evaluate individual companies; he contemplates how technological platforms enable applications that aren't yet obvious. This perspective—informed by his experience building Cue, working at Apple, and mentoring AI startups—provides a framework for identifying opportunity before consensus forms.
The Andromeda Cluster represented a novel investment approach. Rather than simply writing checks, Gross and Friedman provided the infrastructure AI startups needed most. The compute-for-equity model created asymmetric returns: startups that succeeded would generate massive equity value, while the GPU cluster retained value even if individual companies failed.
"What these businesses really need that are getting started in AI today is effectively the equivalent of a white hot oven to run their pizza through," Gross explained. "They need that oven just once or twice to train—to heat up—their basic model and prove to the world that they're good at what they do."
The metaphor captured the AI startup dynamic. Training a competitive large language model might require only a few intensive compute periods. But without access to that compute, even brilliant teams couldn't demonstrate their capabilities. By controlling the "oven," Gross and Friedman became gatekeepers to AI company formation—a position of enormous strategic value.
A consistent theme across Gross's career has been democratization. At Y Combinator, he wanted to "level the playing field for startups to ensure that innovation doesn't get locked up in large companies." At Pioneer, he sought to identify talent "anywhere in the world." With AI Grant and Andromeda, he provided resources that would otherwise concentrate at big tech companies.
The imperative has philosophical roots. Gross spent his youth as an outsider—a religious teenager in Jerusalem who felt disconnected from his peers, a young entrepreneur who left his country to pursue a startup dream. He understood viscerally what it meant to lack access to opportunity.
It also has practical implications. The most valuable companies often emerge from unexpected places and people. By broadening access to resources—funding, compute, networks—Gross increased the probability of discovering exceptional founders. The strategy worked: Pioneer found companies across 50+ countries that traditional VCs never would have encountered.
Gross's involvement with Safe Superintelligence Inc. positioned him in one of AI's most contentious debates: how aggressively should the industry pursue advanced capabilities, and what precautions are necessary?
SSI's founding premise—that safety and capabilities should advance together, not in tension—offered a potential resolution. Rather than slowing AI development for safety concerns, the company aimed to solve safety as a technical problem alongside capability advancement.
"We approach safety and capabilities in tandem," SSI stated, "as technical problems to be solved through revolutionary engineering and scientific breakthroughs."
The framing appealed to investors and researchers who believed that safety-focused deceleration was neither practical nor desirable. If AI development would proceed regardless, better to have safety-conscious organizations at the frontier than to cede leadership to actors with fewer scruples.
Gross's move to Meta complicated this narrative. Meta had faced criticism for releasing powerful AI models (like LLaMA) with relatively permissive licenses, enabling uses that more cautious labs restricted. Working on "superintelligence" at Meta might mean different safety trade-offs than at SSI.
Daniel Gross's career trajectory—from founder to investor back to operator—mirrors a broader shift in the AI industry. The transition from NFDG, a spectacularly successful investment vehicle, to Meta Superintelligence Labs, an operational role, suggests that even the best investors recognize limits to their approach.
In the early stages of a technological revolution, capital allocation creates enormous value. Identifying Uber before others, backing Coinbase when cryptocurrency was speculative, investing in AI companies when machine learning seemed exotic—these decisions generated wealth measured in billions.
But as technology matures, operational execution becomes more valuable than capital deployment. The AI industry in 2025 doesn't lack funding; it lacks people who can build. The talent constraint—not the capital constraint—defines the frontier.
Gross's move to Meta acknowledged this reality. His skills—identifying talent, building teams, managing complex organizations, making strategic decisions under uncertainty—were more valuable applied to building AI systems than to funding AI companies. The same logic drew Friedman from venture investing back into operations.
Gross's journey also illuminates AI power concentration. Over fifteen years, he participated in various attempts to democratize AI—creating YC AI, founding Pioneer, deploying Andromeda, running AI Grant. Yet he ended up at Meta, one of the largest and most powerful technology companies on earth.
The pattern suggests structural forces that resist democratization. Building advanced AI requires resources—compute, data, talent—that concentrate at large organizations. Startups can compete at the margins, in applications and narrow domains, but the foundation model layer increasingly belongs to giants.
Meta Superintelligence Labs assembled a team that few startups could match: Alexandr Wang (Scale AI founder), Nat Friedman (former GitHub CEO), Daniel Gross (Y Combinator partner, SSI co-founder), plus researchers from OpenAI, Anthropic, Google, and Apple. The compensation to assemble this group—potentially billions of dollars—exceeded what most startups raise in their entire existence.
For someone who spent years trying to democratize AI, joining this concentration of power might seem contradictory. But Gross has always been pragmatic. If superintelligence will be built by large organizations regardless, better to be at an organization that might build it responsibly than to critique from the sidelines.
Gross's Israeli origins remain relevant to understanding his approach. He left Israel as a teenager, skipping military service to pursue entrepreneurship. He built a career in Silicon Valley, became a U.S. citizen (Israeli-American by most accounts), and cannot easily return to his homeland due to his unfinished military obligation.
The experience of being an outsider—leaving behind community, country, and expectations—informed Gross's identification with other outsiders. Pioneer explicitly targeted "ambitious outsiders." AI Grant and Andromeda provided resources to founders without establishment connections. Even his investment strategy favored contrarian bets on unconventional founders.
Yet the Israeli tech ecosystem that Gross left has flourished in his absence. The country now produces more AI startups per capita than almost any nation. Israeli engineers occupy senior positions throughout Silicon Valley. The military intelligence units—particularly Unit 8200—that Gross never joined have become legendary talent pipelines for tech companies.
In leaving Israel, Gross traded one exceptional ecosystem for another. He gained access to Silicon Valley's networks, capital, and opportunities. He lost connection to a country increasingly central to global AI development. Whether the trade-off was optimal is ultimately unknowable.
At Meta Superintelligence Labs, Daniel Gross joins an organization with unprecedented resources and ambition. Mark Zuckerberg has committed tens of billions of dollars to AI infrastructure. The team includes some of the industry's most accomplished researchers and operators. The goal—building superintelligent AI systems—represents perhaps the most consequential technological objective ever pursued.
What specifically Gross will build remains unclear. Meta has been characteristically quiet about Superintelligence Labs' internal projects. But given Gross's background in product development, startup acceleration, and talent identification, his role likely involves translating research into applications—bridging the gap between AI capabilities and user-facing products.
The stakes are enormous. If Meta achieves meaningful progress toward superintelligence, it could reshape not just the AI industry but human civilization. The decisions made by people like Gross—about what to build, how quickly to proceed, which safety measures to implement—will have consequences far beyond any individual company's success or failure.
Daniel Gross's fifteen-year journey from Jerusalem teenager to Meta Superintelligence Labs reveals several patterns about the AI industry:
First, speed matters. Gross founded companies, joined companies, invested in companies, and left companies at a pace that would seem reckless in other industries. But AI moves faster than other industries. The technology that defines one moment becomes obsolete within years. Those who hesitate get left behind.
Second, networks compound. Gross's investment success didn't emerge from superior analysis alone. It emerged from relationships with founders, knowledge of emerging companies, and access to deals that others never saw. Each success expanded the network, which created more success.
Third, building beats betting. Despite extraordinary investment returns—4x on a billion-dollar fund in two years—Gross chose to return to building. The observation that the best investors eventually become operators suggests something fundamental about where value accrues in transformational technology shifts.
Fourth, scale wins. Gross spent years trying to democratize AI access. He ended up at Meta. The pattern suggests that in AI, as in earlier technological revolutions, power concentrates despite efforts at distribution. The question isn't whether concentration happens, but who the concentrators will be and how they'll behave.
Several questions about Gross's future remain unanswered:
How will his role at Meta interact with his existing investments? NFDG's portfolio includes companies that might compete with Meta's AI initiatives. The arrangement that enabled his departure—Meta acquiring fund positions without control over portfolio companies—provides some protection, but conflicts seem inevitable.
What happened to Pioneer? The accelerator, which Gross founded in 2018, states on its website that it is "no longer making new investments." The program's alumni have raised substantial capital and built valuable companies, but the organization's current status remains ambiguous.
Will Gross return to entrepreneurship? His career pattern—building Cue, then investing, then building Pioneer, then investing through NFDG, now operating at Meta—suggests periodic returns to company creation. At 33, he has decades of career remaining. Meta may not be his final destination.
How will he navigate AI safety debates within Meta? Gross co-founded a company explicitly dedicated to "safe superintelligence." Meta's approach to AI development, while not reckless, has been more aggressive about open-sourcing powerful models and pursuing commercial applications. Reconciling these perspectives within a single organization presents challenges.
Daniel Gross arrived in Silicon Valley as a teenager with nothing but a Nokia phone, a laptop, and an idea for a search engine. Fifteen years later, he had built and sold a company to Apple, helped define Y Combinator's AI strategy, created a new model for startup acceleration, deployed $100 million in computing infrastructure, invested in companies worth hundreds of billions of dollars, co-founded a $32 billion AI safety company, and joined Meta's effort to build superintelligence.
The trajectory is remarkable not just for its scale but for its consistency. At every stage, Gross positioned himself at the intersection of ambitious people and ambitious technology. He identified talent others overlooked, provided resources others couldn't access, and built networks that multiplied value across hundreds of companies.
His choice to leave investing for operating—to abandon a fund that had quadrupled its money in two years—reveals something important about AI's current moment. We have moved from the era of AI funding to the era of AI building. The people who will shape the technology's future are no longer primarily capital allocators but engineers, researchers, and operators.
TIME magazine named Gross one of the 100 Most Influential People in AI in 2023. The recognition captured his impact as an investor and infrastructure builder. But his most influential years may be ahead. At Meta Superintelligence Labs, with resources that dwarf anything available to startups, Gross will help determine what superintelligent AI looks like—and whether it remains safe.
The outsider from Jerusalem's Orthodox community, who applied to Y Combinator from a military camp and never looked back, now sits at the heart of humanity's most ambitious technological project. Whatever happens next, Daniel Gross will be there to see it—and, more likely, to build it.
"I expect miracles to follow," he wrote when leaving Safe Superintelligence. The question is whether he'll make them happen, and whether the world will recognize them as miracles when they arrive.
## Continue reading - [100 Most Influential People in AI: 2025 Power List](https://digidai.github.io/2025/11/07/silicon-valley-ai-100-most-influential-2025/) - [Yann LeCun: Meta AI Godfather](https://digidai.github.io/2025/11/27/yann-lecun-meta-ai-godfather-world-models-departure-deep-analysis/) - [Jensen Huang: NVIDIA](https://digidai.github.io/2025/11/15/jensen-huang-nvidia-ai-chip-kingmaker-deep-analysis/) - [Sam Altman: OpenAI CEO & AGI Race Leader](https://digidai.github.io/2025/11/08/sam-altman-openai-comprehensive-deep-analysis/)