Photo by Arsyad Basyarudin on Unsplash
Bottom Line
Roughly one in five. As of August 26, 2026, the single most under-discussed number in this rivalry is the one from launch day: Apple Intelligence compatible devices represented approximately 20% of the iPhone installed base when iOS 18 shipped in June 2024, because the feature set required an A17 Pro or M-series chip. Samsung, running the other playbook, pushed Galaxy AI past 100 million devices within its first year after the Galaxy S24 debut in January 2024. Same category, two completely different definitions of "launched."
According to ET Now, whose coverage of this matchup surfaced via Google News, the question is framed as who is winning. That framing is the problem. Reach and refinement are not the same axis, and the platform that wins for a reader depends entirely on which workflow they are trying to close.
What's on the Table
Galaxy AI arrived with the Galaxy S24 series in January 2024 carrying live translation, Circle to Search, and generative photo editing — features aimed at things people already do daily rather than a new assistant paradigm. Samsung then did what Samsung does: spread it. Galaxy AI is now available across flagship, foldable, and select mid-range tiers, riding on a company that shipped approximately 226 million smartphones globally in 2024.
Apple Intelligence went the opposite way. It debuted with iOS 18 in June 2024 built around on-device processing and a privacy-first architecture, and it gated itself behind silicon. iPhone 15 Pro and newer at launch. Nothing older.
Both lean on large language models, but the plumbing differs in a way that matters for anyone thinking about vendor risk. Samsung partners with Google, drawing on Gemini and the Gemini Nano work spreading across the Android ecosystem. Apple runs proprietary models with ChatGPT integration bolted on for the heavier lifts. One is a partnership; one is a stack with a partner attached at the edge.
The Workflow Problem: You Are Buying Availability, Not Features
Here is the scenario that exposes the gap. A three-person consulting team runs client calls in two languages, drafts recaps on the phone between meetings, and cleans up screenshots before they go into a deck. That is live translation, summarization, and generative photo editing — three features both platforms claim.
The feature checklists look nearly identical. The availability math does not.
Run the numbers from the research side by side. Samsung shipped roughly 226 million phones in 2024 and got Galaxy AI onto more than 100 million devices in year one — call it a 44% attach rate against a single year of shipments, and that understates it, because Galaxy AI also landed on devices already in pockets. Apple's roughly 20% installed-base coverage at iOS 18 launch is measured against the whole iPhone base, not one year of sales. So the honest comparison is not "100 million versus 20 percent." It is: Samsung optimized for how many people could touch the feature this quarter, Apple optimized for how well it ran on the hardware it allowed in.
Chart: Galaxy AI's first-year device reach against Samsung's 2024 shipments, versus Apple Intelligence's share of the iPhone installed base at iOS 18 launch. Units are deliberately not normalized — shipments, device count, and installed-base share measure different things, which is exactly why single-source "who's winning" headlines mislead. Figures as reported in the research cited above, current as of August 26, 2026.
The skeptic's pushback is fair and worth naming: raw device count is a vanity metric. A feature enabled on 100 million handsets that half the owners never open is worth less than a tightly optimized feature used daily by a smaller cohort. True. But it cuts both ways — an exquisitely optimized summarizer is worth exactly zero to someone holding an iPhone 14. Availability is the floor under every other argument.
Who Wins Under Which Condition
Analysts covering this space have converged on a point that gets lost in feature-count journalism: the smartphone AI race is less about capability lists and more about ecosystem integration and whether users trust the privacy model. That reframing decides the matchup cleanly.
Samsung wins on mixed-hardware teams. If a small team is running a spread of devices — a flagship, a foldable, two mid-range handsets — Galaxy AI's tier spread means everyone gets the same live translation and Circle to Search behavior. Nobody is left out of the shared workflow. This works for a team of 3 and, notably, still works at 30, because the compatibility line is drawn generously.
Apple wins on regulated or sensitive data. The on-device-first architecture is a genuine compliance argument, not marketing. With privacy regulation in the EU and US actively shaping how vendors split on-device versus cloud processing, "this text never left the handset" is a sentence a legal team can work with. Apple's control over both silicon and OS also yields tighter optimization, which is the real payoff of the hardware gate.
Neither wins on model independence. This is the second-order consequence nobody markets. Samsung's AI capability is partly Google's roadmap — Gemini and Gemini Nano across Android. Apple's heaviest queries route to ChatGPT integration. In both cases, a user's daily workflow sits downstream of a third party's model deprecation schedule. The same cost-and-dependency logic that AI Trends traced in enterprise Claude model selection applies here: the model behind the button can change, and the button will not tell you.
Meanwhile Qualcomm and MediaTek have accelerated AI-capable chipset work to feed these initiatives, which quietly raises the hardware floor for everyone. The practical translation: the phone that runs today's AI features comfortably may be the one that struggles in two generations, on both platforms.
The Real Limit Nobody Markets
Three limits deserve a place in any buying decision, and none of them appear in a keynote.
Apple Intelligence required an A17 Pro or M-series chip at launch, which is why coverage sat near 20% of the installed base. Before budgeting for an AI-driven workflow, verify the silicon in the devices already deployed. A feature that requires a hardware refresh is a hardware purchase wearing a software label.
Not the demo. Pick the single task performed most — call recaps, translated messages, photo cleanup — and run it ten times on both platforms if possible. Reviews and benchmarks show the gap between platforms narrows sharply on routine tasks and widens on edge cases. Routine tasks are what a workflow is made of.
The on-device versus cloud split is the whole privacy story, and it varies feature by feature within each platform. For anyone handling client data, document which specific features stay local. Regulatory pressure in the EU and US is already reshaping these boundaries, so a policy read from last year may no longer describe current behavior.
The Verdict
Our read: framing this as a winner-take-all fight misreads what both companies are doing. Samsung optimized for reach and got it — 100 million-plus devices in year one is a distribution achievement, and generative photo editing on a mid-range handset is a more consequential product decision than another point of benchmark performance. Apple optimized for a defensible privacy posture and tighter hardware-software integration, and accepted a roughly 20% starting base to get it.
On balance, the more likely outcome is convergence rather than a knockout. Samsung's near-term advantage in feature reach is real but not durable — Apple's installed-base coverage rises mechanically with every upgrade cycle, no engineering required. The more interesting variable is trust: whichever platform can credibly say where a user's data goes, feature by feature, ends up owning the professional segment. That is a documentation problem as much as an engineering one, and neither company has solved it yet.
Still rough on both sides, then. Buy for the hardware you can actually deploy today, and treat the AI features as a tiebreaker rather than the reason.
Frequently Asked Questions
Is Apple Intelligence available on older iPhones?
Not at launch. Apple Intelligence required an A17 Pro or M-series chip when it debuted with iOS 18 in June 2024, which limited it to iPhone 15 Pro and newer models. That hardware requirement is why compatible devices accounted for roughly 20% of the iPhone installed base at launch.
Which phones get Galaxy AI features besides the flagship models?
Galaxy AI is available across multiple Samsung device tiers, including flagships, foldables, and select mid-range models. That broader compatibility is the core reason Galaxy AI reached over 100 million devices within its first year after the January 2024 Galaxy S24 launch.
Does Galaxy AI or Apple Intelligence handle privacy better?
Apple Intelligence was built around on-device processing as an explicit design principle, which is a stronger default privacy position. But the honest answer is feature-by-feature: both platforms route some requests to the cloud, and Samsung's Google partnership and Apple's ChatGPT integration both involve third-party model providers. Verify per feature rather than per brand.
Whose AI models actually power these features?
Samsung partners with Google, leaning on Gemini and the Gemini Nano work rolling out across the Android ecosystem. Apple uses proprietary models with ChatGPT integration for heavier requests. Neither platform is fully self-contained, which means both carry some exposure to a partner's model roadmap.
Disclaimer: This article is editorial commentary based on publicly reported information and does not constitute product testing, purchasing advice, or financial advice. No independent hands-on evaluation of these platforms was conducted for this piece. Device availability, feature sets, and processing policies change frequently — verify current specifications with the manufacturer before purchase. This site may earn affiliate commissions from links to products mentioned; no such relationship influenced the analysis above. Research based on publicly available sources current as of August 26, 2026.