2026 Predictions: Your Face is AI’s Next Battleground
2026 Predictions: Your Face is AI's Next Battleground
As we approach a new year, it's time to reflect on the lessons from the past twelve months and formulate the outlook for the next twelve. 2025 was an incremental year for spatial computing, which continues its gradual ascent toward mainstream traction.
2025 Highlights
2025 saw significant advancements in the convergence of XR (Extended Reality) with AI (Artificial Intelligence), inflections in non-display AI glasses, the rise of video display glasses, and the unveiling of the long-awaited Meta Ray-Ban Display glasses. Meanwhile, roadmap signals emerged from players ranging from Snap to Apple.
Form-Factor Divergence and Diversification
All these approaches – video passthrough AR (Augmented Reality), optical see-through AR, and non-display smartglasses – represent form-factor divergence and diversification. This is a good thing, as XR should include varied formats that are purpose-built and use-case-driven – a key trend in 2025.
Prediction 2: Your Face is the Next AI Battleground
With that backdrop, what will 2026 look like in spatial computing? Aligned with the broader predictions of our research arm, ARtillery Intelligence, we've devised 5 predictions for 2026. We'll break them down weekly, continuing here with #2: Your face is the next AI battleground.
AI's Unrecognized Value in the Physical World
One of AI's next frontiers will be your face. This is part of the broader principle of AI's unrecognized value in the physical world. Most of AI's outpouring of investment and excitement over the past two years has been confined to digital domains.
For example, Generative AI and large language models mostly operate within the reaches of the web, appverse, and things that happen on your screen(s). Greater value awaits in the physical world. Once AI can tap into and train itself on all that data, the value creation will be immense.
Bringing AI to Your Face
To be fair, this isn't a new concept, covered in fields ranging from IoT (Internet of Things), to digital twins, to Nvidia's Omniverse. But when it comes to high-value consumer-facing AI, the fact remains that most of it involves data that's been trained on/from the web, with interfaces that involve 2D screens.
Bringing this back to your face, if AI can see what you see, it can be a more powerful tool to gain contextual awareness and elevate intelligence. That is the best way to augment human intelligence – more so than sensors that reside elsewhere, such as on your wrist or in your pocket.
The Opportunity for Your Face to Not Just Be the Data Collection Point
Importantly, the opportunity is for your face to not just be the data collection point but the consumption point. That intelligence gathered, trained, processed, and returned can take shape in AR that informs you about your surroundings. As we've learned, that can be visual or audible.
Who's Primed to Win the Coveted Spot on Your Face?
The remaining question is who's primed to win that coveted spot on your face. Meta is ahead in terms of progress in AI smart glasses. But it could have trust issues with data collection. Apple has ample ammunition to catch up to Meta – and fewer trust issues – but has stumbled in AI.
Google sits between them with the best of both worlds, given fewer trust issues (albeit some), and greater AI. The latter is Google's ace in the hole, considering that Gemini will be sewn into Android XR – thus tapping into Google's extensive and time-tested knowledge graph.
Snap's Prime Position in 2026
Meanwhile, Snap will be in a prime position in 2026, given that consumer Spectacles' AR chops will exceed all the above. In other words, consumer AR glasses in 2026 will offer "flat AR," while Spectacles offer dimensional (SLAM) AR – so far limited to enterprise devices.
Ray-Ban Meta Smartglasses (RBMS) and Meta Ray-Ban Display Glasses (MRBD)
Looking back on 2025 predictions, we see that Ray-Ban Meta Smartglasses (RBMS) were the world's introduction to faceworn AI at scale. 2026 unit sales will exceed 4.8 million – a substantial but decelerated growth rate from 2025 as the market approaches a saturation point relative to its current demand ceiling.
Meta Ray-Ban Display Glasses (MRBD) – carrying the same principle but evolved into visuals – will sell far fewer units due to its price tag, coming in at 30,000 units for the year. In both cases, Meta will warm up the market, leading up to a 2027 inflection (covered in the next prediction).
Conclusion
We'll pause there and circle back next week with another 2025 prediction. Meanwhile, see the full report. And to calibrate our aptitude and track record in projecting XR market outcomes, see our recent article evaluating last year's predictions.
# Example code to illustrate the concept of AI's unrecognized value in the physical world
import numpy as np
# Define a function to simulate the value creation in the physical world
def simulate_value_creation(data):
# Simulate the training of AI on the data
ai_model = train_ai_model(data)
# Simulate the value creation in the physical world
value = ai_model.predict(data)
return value
# Define a function to train the AI model
def train_ai_model(data):
# Simulate the training of the AI model
ai_model = np.random.rand(10, 10)
return ai_model
# Define a function to simulate the data collection
def simulate_data_collection():
# Simulate the data collection
data = np.random.rand(10, 10)
return data
# Run the simulation
data = simulate_data_collection()
value = simulate_value_creation(data)
print(value)
This code illustrates the concept of AI's unrecognized value in the physical world by simulating the value creation in the physical world using a trained AI model. The code defines three functions: simulate_value_creation, train_ai_model, and simulate_data_collection. The simulate_value_creation function simulates the value creation in the physical world by training an AI model on the data and predicting the value. The train_ai_model function simulates the training of the AI model, and the simulate_data_collection function simulates the data collection. The code runs the simulation by calling the simulate_data_collection function to collect the data, and then calling the simulate_value_creation function to simulate the value creation in the physical world.
Source: https://arinsider.co/2025/11/19/2026-predictions-your-face-is-ais-next-battleground/
