In another bold move underscoring its ambitions to dominate the artificial intelligence (AI) frontier, Meta has hired Apple’s head of AI models, Ruoming Pang, marking a high-profile defection that underlines the escalating talent war among Silicon Valley’s tech giants.

Pang’s departure, first reported by Bloomberg, comes at a critical time for Apple as it struggles to keep pace with rivals in developing cutting-edge generative AI systems.

Pang led Apple’s in-house foundation models team, a group of over 100 engineers responsible for training the neural networks powering Apple Intelligence—the company’s newly launched suite of AI features. These include on-device functions such as Genmoji, Priority Notifications, and Mail summarization, all slated to roll out with iOS 18 later this year.

However, sources familiar with Apple’s AI efforts say the company’s models have failed to keep up with rivals like OpenAI, Anthropic, and even Meta itself. Apple has reportedly considered licensing external models to bridge the gap, including options from OpenAI and Anthropic, particularly for its upcoming Siri overhaul.

Pang’s departure may be just the beginning. People with knowledge of Apple’s AI division say other senior engineers could soon follow him out the door, casting further doubt over the company’s internal capacity to lead in the rapidly evolving AI space.

Meta’s Superintelligence Labs: A Magnet for Talent

Pang will now join Meta’s Superintelligence Labs, the elite AI division launched by CEO Mark Zuckerberg to accelerate the company’s long-term AI roadmap. Superintelligence Labs has become a destination for elite researchers and engineers, recruiting talent from OpenAI, Google DeepMind, and now Apple.

Meta’s Superintelligence Labs is helmed by Alexandr Wang, former CEO of Scale AI, a data labeling startup Meta recently acquired through a $14.3 billion investment. Wang now serves as Meta’s Chief AI Officer. Alongside him is Nat Friedman, the former CEO of GitHub, who will oversee Meta’s AI product strategy and applied research. The team also includes Daniel Gross, Friedman’s longtime business partner and the co-founder of Safe Superintelligence, a high-profile startup that rebuffed Meta’s acquisition offer earlier this year.

The team’s goal is to develop next-generation AI systems capable of multimodal reasoning, natural conversation, and even autonomous decision-making across Meta’s ecosystem.

Meta has reportedly offered compensation packages ranging from $10 million to $100 million for key hires, underlining just how serious Zuckerberg is about building what he’s called a “frontier AI platform”—an end-to-end system that will not only power Meta’s social platforms but also lead in enterprise and developer tools.

For Pang, the move represents both a leap into a more advanced AI environment and a chance to help shape the architecture of Meta’s AI offerings from the ground up.

Pang’s exit also reflects broader cracks in Apple’s AI strategy. While the company has prioritized privacy and on-device intelligence, that approach has limited its ability to deliver the kind of rich, real-time generative AI experiences users now expect. Meanwhile, competitors are racing ahead with large-scale cloud-based models, pushing updates at a blistering pace.

Meta’s poaching of Pang is just the latest in a long line of aggressive hires that illustrate the company’s intent to dominate AI. Earlier this year, it recruited top scientists from Google’s DeepMind, secured researchers from Safe Superintelligence (SSI), and added ex-OpenAI contributors to its AI safety and alignment units.

As Meta assembles what some in the industry have described as the “Avengers of AI,” Apple finds itself facing mounting pressure—not just to ship features, but to retain the talent capable of building them.

While Apple Intelligence will debut later this year in a suite of consumer-facing features, the company’s reliance on third-party AI providers for core capabilities suggests it is still playing catch-up. In contrast, Meta’s Superintelligence Labs is becoming a self-contained AI powerhouse, poised to lead innovation and influence the next wave of AI products across platforms and industries.

Pang’s jump from Apple to Meta may appear as a single hire—but it represents a seismic shift in the battle for AI supremacy. And with Meta aggressively stacking its AI roster, more shakeups are likely to follow.

The landscape of global innovation is shifting at an unprecedented pace, and recent trends at Tekedia Capital highlight a critical divergence. Our latest cohort of 18 companies, invested in just three months ago, is demonstrating the fastest revenue growth we’ve ever observed, surpassing even the strong performance of our October 2024 class. This earlier class included a remarkable company that achieved a $10 million Annual Recurring Revenue (ARR) within four months of its inception and is now on track for $100 million within its first year, having recently secured $36 million in funding.

This data leads to a stark conclusion: the innovation gap between developing and developed nations is not merely present, but rapidly widening to an asymmetrical degree. The defining characteristic of leading companies today is “AI-nativity.” These are not just firms using artificial intelligence, but rather businesses fundamentally built upon AI, regardless of their industry. For instance, the most successful insurance companies of tomorrow will likely be AI companies that happen to offer insurance products, not traditional insurance firms merely integrating AI. Similarly, a premier online tutor like ChatGPT isn’t an edtech company that uses AI; it’s an AI company providing educational services.

When we contrast these emerging, AI-native enterprises with many African startups, a troubling pattern emerges. The gap in the pace of innovation and the value delivered to users is expanding. There’s a tangible risk that many African Software-as-a-Service (SaaS) companies could face overnight disintermediation.

This observation is not an alarmist prediction, but a conclusion drawn from reviewing the financials of dozens of companies across five continents in Tekedia Capital. In this rapidly accelerating era of AI, many African startups are struggling to keep pace, and a significant portion of our SaaS companies could potentially disappear by 2027. Urgent action is required. We must fundamentally rethink our approach to building businesses to ensure Africa can compete and win in this new global landscape. I have called the landscape the accelerated society era and we must find our level there productively.

The Accelerated Society: A New Era of Disruption and Integration

Block CEO and Twitter co-founder Jack Dorsey has launched Bitchat, a radical new messaging platform that operates completely off the internet, signaling a major push toward decentralized, surveillance-resistant communication.

Announced over the weekend, the beta version of the app is now available on Apple’s TestFlight, with a white paper published on GitHub detailing the technical foundations of what Dorsey describes as a “personal experiment” in alternative connectivity.

Unlike mainstream messaging platforms such as Meta’s WhatsApp, Messenger, or Telegram, Bitchat does not require a phone number, email, or any central server to function. The app works entirely over Bluetooth mesh networks, allowing users to send encrypted messages between nearby devices — even in the absence of Wi-Fi, mobile data, or internet infrastructure.

A Challenge to Centralized Messaging Giants

Dorsey’s latest project is already being seen as a potential rival to WhatsApp, particularly in regions where internet shutdowns, government surveillance, or data privacy concerns are prevalent. While WhatsApp remains the dominant player in mobile messaging — with over 2 billion users globally — it requires user registration through personal identifiers like phone numbers, stores data on cloud servers, and has faced scrutiny over metadata collection and integration with Meta’s broader advertising ecosystem.

Bitchat, in contrast, is a privacy-first tool that doesn’t track users or store any messages outside of devices. Its peer-to-peer architecture ensures that conversations never touch centralized infrastructure, and messages are ephemeral by default, disappearing once delivered or after a set time.

How Bitchat Works

Bitchat leverages Bluetooth mesh networking and store-and-forward technology, enabling messages to hop across multiple devices like digital whispers in a crowd. Even if the recipient is not in range, another user’s phone can hold and later forward the message, making the system resilient in disaster zones, protest areas, remote regions, or under authoritarian censorship.

The app also includes support for group chats or “rooms,” which can be named using hashtags and locked behind passwords. In future updates, WiFi Direct support is expected, expanding the app’s operational range and speed while maintaining the infrastructure-free design.

This setup mirrors technology used in Bridgefy, a Bluetooth messaging app that became popular during the Hong Kong protests of 2019, and in Briar, another off-grid messaging app used in authoritarian states and emergencies. But Bitchat’s simple interface, strong encryption focus, and Dorsey’s influential backing could allow it to break through to a broader audience.

Aligning with Dorsey’s Push for Decentralization

The launch follows Dorsey’s continued advocacy for decentralized platforms. After stepping away from Twitter, he backed Bluesky, a federated social media protocol, and supported Damus, a Nostr-based app that runs on open standards with no central authority.

Bitchat is an extension of this mission. Dorsey called it an experiment in “Bluetooth mesh networks, relays, store-and-forward models, message encryption models, and a few other things” — essentially, a testbed for off-grid communication that could upend how people connect in restrictive or crisis environments.

Implications for Global Messaging and Security

Dorsey’s move comes at a time when governments are clamping down on encrypted communication, and tech platforms are facing increasing demands to hand over user data. In this climate, apps like Bitchat are not just technological experiments — they are statements of digital autonomy.

The potential to rival WhatsApp, especially among users concerned with privacy or living under regimes prone to surveillance or internet control, is substantial. While WhatsApp has end-to-end encryption, its reliance on centralized servers and phone-based identifiers creates attack surfaces that Bitchat aims to eliminate entirely.

Bitchat is still in beta and available only to iOS testers for now, but its ambition is clear. With no central database, user accounts, or app-based identity, it embodies one of the most radical visions yet for peer-to-peer communication.

In an era of growing concerns over state surveillance, tech censorship, and infrastructure failure, Jack Dorsey’s Bitchat is carving out a space where freedom of speech and privacy do not depend on connectivity — or on big tech platforms.