AI race reboots: giants double down as startups carve new niches.

Apple has just rolled out iOS 27, one of its most significant software updates in years ([1]). The update transforms Siri from a relatively limited assistant into a full-fledged generative AI engine built directly into the iPhone’s operating system. Siri can now leverage broad world knowledge, personal context, on-screen understanding, and perform complex multi-step actions across apps ([2]). The catch: only newer devices (iPhone 15 Pro and up) unlock all advanced features, reflecting how AI capabilities increasingly tie into hardware upgrades ([3]).

Strategically, this move marks Apple’s aggressive pivot to catch up with competitors like OpenAI, Google, and Microsoft, who surged ahead in the generative AI boom ([4]). By baking AI deeply into iOS, Apple immediately deploys an upgraded Siri to hundreds of millions of devices ([5]) – a distribution heft that no startup or single-product rival can match. Early indications suggest Apple might monetize the most resource-intensive AI features over time, signaling a potential new subscription revenue stream for advanced Siri capabilities ([6]) ([7]).

In a twist, Apple may not be going it alone in the AI model race. Developers digging through iOS 27 code found a new “Model Delegation” framework that would let Siri hand off requests to third-party AI models like Anthropic’s Claude or OpenAI’s GPT series ([8]). In fact, hidden interfaces allow replacing Apple’s own cloud-based Siri model entirely with an external model for certain tasks ([9]). This suggests Apple’s strategy is to turn Siri into an orchestrator of AI services – maintaining control over the user experience and device integration while tapping whichever AI engine is best for the job. It’s a pragmatic path that could give Apple a unique platform advantage, effectively making iOS a high-scale gateway for multiple AI ecosystems ([10]). Combined with regulatory pressure from the EU’s Digital Markets Act to open up core functionality like Siri to outside integration ([11]), Apple’s new approach could redefine platform competition: rather than build the most powerful model, become the default hub through which consumers access AI.

While Apple was reshaping its product strategy, SoftBank was rewriting the playbook on AI investment. The Japanese conglomerate secured an $11.87 billion loan from a consortium of about 20 banks – overshooting its initial $10 billion target ([1]) – to pump even more money into OpenAI. SoftBank has now committed roughly $64.6 billion in total funding to OpenAI and expects to reach that figure by next month, which would give it an estimated 13% ownership of the AI lab ([2]). This ranks among SoftBank’s largest bets ever, effectively tying a significant part of its future to a single AI platform leader.

The sheer scale of SoftBank’s wager underscores a growing conviction in the winner-takes-most dynamics of AI. If OpenAI maintains its technical edge and ecosystem, SoftBank stands to wield substantial influence – and reap enormous rewards – across the AI economy. However, this confidence isn’t universally shared by public markets. News of prominent AI executives calling for a slowdown in “frontier” model development sent semiconductor stocks sliding worldwide and contributed to SoftBank’s own stock plummeting 13% – its worst drop since July ([3]). In contrast, banks readily over-subscribed SoftBank’s loan, signaling that private capital remains eager to fuel AI’s leaders despite short-term risk signals ([4]).

For C-level strategists, the takeaway is twofold. First, expect a continued concentration of capital and capability around a few AI powerhouses – startups that can credibly lead in foundation models or essential AI infrastructure will attract outsize investment. Second, the divergence between public market jitters and private funding enthusiasm means there’s perceived long-term value in AI that may not yet be reflected in quarterly stock prices. Companies must balance short-term caution with long-term positioning: those who can align with, invest in, or develop platform-level AI capabilities stand to gain a durable competitive edge, even as the broader market sorts out how to value AI’s transformative potential.

Google’s latest move highlights the fierce competition for AI talent and technology – especially in areas where incumbents feel vulnerable. On September 14, Google confirmed a deal valued at over $1.5 billion to effectively acqui-hire Mechanize, a San Francisco AI startup with barely a dozen core researchers ([1]) ([2]). Mechanize’s co-founder and CEO Tamay Besiroglu (who previously helped found AI evaluator Epoch) has joined Google’s DeepMind, bringing along more than a dozen top specialists in using AI agents to write and improve code ([3]). Notably, Mechanize had raised only about $9 million earlier this year at a $500 million valuation ([4]) – a remarkable leap that underscores how critical AI coding capabilities have become in the eyes of Big Tech.

Behind this expensive talent swoop is a strategic urgency. Code generation and software automation are now seen as a high-stakes frontier in AI, and by Google’s own admission, coding has been a relative weak point for its models compared to OpenAI’s Codex and GitHub Copilot or Anthropic’s Claude Code ([5]) ([6]). Rather than cede this ground, Google is effectively buying the expertise it needs. The structure of the Mechanize deal – hiring the team and licensing its technology, without an outright acquisition – also reflects a playbook to avoid antitrust scrutiny ([7]). Google ran a similar play in 2025 by poaching talent from AI coding startup Windsurf after OpenAI showed interest ([8]) ([9]). The pattern is clear: nimble AI research teams tackling key problems are being snapped up by giants long before they can grow into full-fledged competitors.

For industry players, Google’s move sends a sobering competitive signal ([10]). If your strategy relies on a unique AI breakthrough – especially one that addresses an incumbent’s pain point – expect the big platforms either to replicate it quickly or to open their checkbooks. From a proactive standpoint, established companies should continually assess where their AI capabilities lag and consider strategic acquisitions or partnerships to prevent a disadvantage. At the same time, startups in the AI tools space may find that the fastest path to market impact (and monetization) is via an early acquisition, not independent growth. The downside is a thinning field of independent AI providers, which could funnel more power to the largest tech firms over time.

Amid the big-company maneuvers, AI-native startups are staking out opportunities in areas that remain unsolved by the tech giants. One striking example is Nuance Labs, a Seattle startup founded by three ex-Apple researchers, which just announced a $50 million Series A round led by Lightspeed Venture Partners (with participation from Accel and even NVIDIA’s venture arm) ([1]). Nuance’s mission is to make AI fundamentally better at face-to-face conversation – teaching AI “avatars” to not only generate speech but also to interpret and respond to human non-verbal cues in real time ([2]). The company is developing a full-duplex AI model that can listen and speak simultaneously, enabling more natural timing, tone, gaze, and responsiveness during dialogue. In other words, they want to eliminate the robotic pauses and blank stares that plague today’s virtual agents ([3]).

Strategically, this is about unlocking AI’s next user interface. If Nuance succeeds, it could upend industries from customer service and telehealth to virtual reality, where lifelike digital assistants and companions have long been envisioned. Big tech companies have their own virtual assistant initiatives – Apple’s new Siri, Meta’s avatar efforts, Zoom’s AI meeting features – but even the most advanced lack the fluid, human-like exchange that Nuance is targeting. By securing "foundation-model scale" funding to solve these interaction challenges ([4]), specialized startups like Nuance aim to leap ahead in a niche that the larger players have yet to perfect. For business leaders, it’s a reminder that some AI competitive advantages will come from quality of experience, not just raw intelligence. A company that can humanize AI interactions may redefine customer expectations and seize opportunities in the next wave of AI-driven products or services.

Recent funding patterns indicate that the era of blank-check investing in any startup with “AI” in its pitch is fading. On September 14 alone, three AI-enabled startups – Hong Kong fintech Qupital, Sweden’s Tandem Health, and Dutch cybersecurity firm Fortaegis – announced roughly $450 million in combined financing ([1]). Qupital raised a $300 million mix of equity and credit to fuel AI-powered lending for e-commerce merchants, Tandem Health secured $100 million to build an AI-driven operating system for clinics, and Fortaegis landed $50 million to embed security directly into AI hardware. Each of these ventures marries artificial intelligence with hard-to-replicate assets: proprietary real-time data or deep domain expertise, highly regulated customer environments, or innovation at the infrastructure level ([2]) ([3]).

The broader message from investors is clear: AI by itself isn’t a defensible moat ([4]). Instead, value is emerging where AI is paired with something uniquely hard to copy – be it exclusive datasets, tight integration into critical workflows, or technical barriers like custom silicon. As one tech funding analyst put it, “AI is becoming less interesting as a standalone label and more valuable when paired with scarce data, regulated access, financial infrastructure, [or] entrenched workflows” ([5]). In practice, that means capital is now flowing to companies that use AI to fortify an existing strategic advantage, rather than those just wrapping generic models in a new app.

For executives, this marks a shift in how to approach AI initiatives and partnerships. It’s no longer sufficient to sprinkle machine learning features on top of a product and call it a strategy. The competitive gains will more likely come from using AI to enhance core strengths – whether it’s a wealth of proprietary data, a strongly defended market position, or control over a key distribution channel. Companies should evaluate where they have—or can build—such "AI plus X" advantages in their business. Those that effectively combine AI with their unique assets or capabilities will create more enduring competitive moats, while those without such differentiation risk being outpaced by the next wave of smarter, more deeply integrated competitors.