The Sensory Internet

The Bandwidth Mismatch

Human sensory processing operates at approximately 11 million bits per second across all channels. Vision accounts for roughly 10 million. Touch accounts for approximately 1 million. Hearing accounts for approximately 100,000. Smell and taste account for the remainder.

The current internet communicates through two channels: vision (text, images, video) and hearing (audio, voice). This captures approximately 92% of sensory bandwidth by raw throughput but misses the embodied, spatial, and tactile dimensions that define physical experience. A video call transmits faces and voices but not the handshake, the room temperature, the spatial geometry of the meeting, or the subtle pressure of a pen on paper.

The mismatch matters commercially and experientially. A surgeon learning a procedure from a video receives visual information but no haptic feedback: the resistance of tissue, the pressure required for incision, the texture differences between healthy and pathological tissue. A remote worker on a video call receives visual and auditory information but no spatial presence, the embodied sense of being in the same room. An online shopper sees a product image but cannot feel the fabric weight, the surface texture, or the fit.

Each missing sensory channel represents a fidelity gap between digital and physical experience. The sensory internet is the project of closing these gaps, channel by channel, until digital experiences approach the full bandwidth of physical ones.

Haptic Technology: Teaching the Internet to Touch

The haptic technology market reached $4-12 billion in 2025 (estimates vary by scope: hardware-only vs. full ecosystem), with projections reaching $8-14 billion by 2030. Growth is driven by gaming, automotive HMI, surgical simulation, and XR applications.

The technology is advancing through three distinct generations, each expanding the vocabulary of digital touch:

Generation 1: Vibration (1990s-present). Eccentric Rotating Mass (ERM) motors. The vibration motor in your phone. Binary feedback: on or off, one intensity. This is what most current devices use for notifications and basic game feedback. Cheap, reliable, expressively limited.

Generation 2: Linear actuators (2015-present). Linear Resonant Actuators (LRAs). Used in Apple's Taptic Engine and Sony's DualSense controller. These provide variable intensity, frequency, and waveform control, enabling textures, clicks, and directional feedback. The DualSense's adaptive triggers simulate the tension of drawing a bowstring or the resistance of pressing a car brake. Apple's Taptic Engine makes a flat glass surface feel like it has physical buttons.

Generation 3: Force feedback and surface haptics (2020s-emerging). Ultrasonic arrays that project focused pressure in mid-air (Ultraleap), electroactive polymers that change shape and stiffness when voltage is applied, and microfluidic arrays that create localized temperature and pressure sensations. These technologies enable the simulation of touching objects that exist only in digital space.

The latency requirement for haptic realism is approximately 23 milliseconds: the threshold below which the brain perceives touch as instantaneous. Current 5G networks achieve 10-20ms latency under ideal conditions. Earlier network generations could not consistently meet this requirement, which is why haptic interaction over networks has been impractical until the current generation of mobile infrastructure.

Spatial Computing: The Room Replaces the Screen

Apple Vision Pro, released in February 2024, defined the "spatial computing" category: a paradigm where digital content is placed in, and interacts with, the physical environment rather than being confined to a rectangular screen.

The interface operates through three input modalities: eye tracking (where you look), hand gestures (what you do with your hands), and voice (what you say). No controllers, no mouse, no keyboard for primary interaction. The device tracks the user's environment through cameras and LiDAR, renders digital objects with real-world lighting and shadows, and allows natural interaction through gaze and gesture.

The market reception has been instructive:

• Apple Vision Pro: IDC reports approximately 390,000 units shipped in 2024, dropping to roughly 45,000 in 2025. The high price point ($3,499), bulky form factor, and limited native app library constrained mass adoption. Apple has pivoted toward enterprise and luxury retail use cases.

• Meta Quest: Meta maintains 60-80% of VR/MR hardware sales through aggressive pricing (Quest 3 at $499). In 2025, Meta reported its all-time highest number of unique active users. Over 100 apps on the Meta Horizon Store generated more than $1 million in gross revenue.

• Smart glasses pivot: By late 2025, the industry shifted toward lighter, more wearable smart glasses (Ray-Ban Meta series) as a potentially more viable mass-market path than bulky headsets.

The broader spatial computing market was valued at approximately $149-182 billion in 2024 and is projected to exceed $800 billion by 2032 (various analyst estimates). The distinction matters: most of this value is in enterprise applications (digital twins, industrial simulation, training), not consumer headsets.

The Missing Piece: Haptic Presence

The gap in spatial computing is haptics. Current spatial interfaces are visual-gestural: you see digital objects and manipulate them with hand movements, but there is no physical resistance. Picking up a virtual object provides no weight. Pressing a virtual button provides no click. This gap between visual realism and tactile absence is the primary barrier to the "presence" that spatial computing promises.

Companies working on this gap:

• Ultraleap: Mid-air haptics using focused ultrasound. You can "feel" virtual buttons and textures without wearing gloves.
• HaptX: Force-feedback gloves that provide realistic grasping sensations. Used by Nissan, SAP, and the U.S. Air Force for training.
• Meta Reality Labs: Developing prototype haptic gloves with pneumatic actuators that simulate pressure and texture.

Brain-Computer Interfaces: Bypassing the Senses

The most direct solution to the sensory bandwidth problem is bypassing physical senses entirely.

Neuralink implanted its first human brain-computer interface (BCI) in January 2024. The N1 implant, placed in the motor cortex, enabled a quadriplegic patient to control a computer cursor through thought alone: moving the cursor, clicking, scrolling, and typing at speeds approaching those of able-bodied users.

The BCI market reached approximately $2.3-3 billion in 2025, with projections of $13-15 billion by 2035. The competitive landscape is bifurcating:

Invasive BCIs (surgical implantation):

• Neuralink: N1 implant, motor cortex placement, ~1,024 electrodes. PRIME clinical trial expanding across US, Canada, and UK. Received FDA Breakthrough Device designation for "Blindsight" (vision restoration) and speech restoration. Valued at approximately $9-10 billion (mid-2025).
• Synchron: Stentrode, an endovascular BCI inserted through blood vessels (no open brain surgery). FDA Breakthrough Device designation. Lower surgical risk but fewer electrodes.
• Precision Neuroscience: "Layer 7" cortical interface that sits on the brain surface. Less invasive than Neuralink, more electrodes than Synchron.
• Paradromics: Connexus Direct Data Interface, high-bandwidth cortical implant targeting speech and communication restoration.

Non-invasive BCIs (external sensors):

• EEG-based headsets: Consumer devices from companies like Emotiv and Muse. Lower resolution, no surgery, used for meditation, gaming, and basic cursor control.
• fNIRS (functional near-infrared spectroscopy): Measures blood oxygenation through the skull. Slower than EEG but higher spatial resolution. Used in research and emerging consumer applications.

The current medical applications are focused on restoration: enabling paralyzed patients to interact with digital devices, restoring communication for patients with ALS, and potentially restoring sensory input for patients with spinal cord injuries.

The longer-term trajectory (10+ years) is bidirectional neural interfaces that provide communication in both directions: brain-to-device (control) and device-to-brain (sensation). This would create a fundamentally new medium. Information would be transmitted not through eyes and ears but directly into sensory cortex. Touch, temperature, spatial awareness, and potentially novel sensory modalities that have no biological equivalent could be delivered directly.

The Sensory Internet Stack

The sensory internet is not a single technology. It is a stack of technologies that, layered together, progressively close the gap between digital and physical experience.

The Presence Threshold

The commercial and experiential question is whether these technologies can cross the "presence threshold": the point at which a digital experience is perceptually indistinguishable from a physical one for practical purposes.

Current estimates place this threshold at a combination of requirements:

• Visual resolution: exceeding 60 pixels per degree (achieved by Vision Pro at ~34 ppd for central vision; full threshold requires ~60 ppd which is approaching human retinal limits)
• Haptic latency: below 23ms (achievable on 5G networks under ideal conditions)
• Spatial audio: head-related transfer functions with individualized calibration (standard in spatial computing devices)
• Motion-to-photon latency: consistent sub-20ms (the primary contributor to motion sickness in VR; Quest 3 achieves ~12ms)
• Frame rate: minimum 90fps for comfort, 120fps for full presence (achievable on current hardware)

No current system meets all requirements simultaneously across all sensory channels. But each requirement is individually achievable with existing or near-term technology. The integration challenge, combining all sensory modalities with consistent low latency in a device that is comfortable for extended use, is an engineering problem, not a physics problem.

The Commercial Implications

The sensory internet does not merely add features to the existing internet. It creates new categories of experience that are impossible with screens and speakers alone.

Surgical training. A surgeon practicing on a haptic-enabled simulation feels the resistance of tissue, the feedback of instruments, and the spatial geometry of the operating field. The transfer learning from haptic simulation to physical surgery is significantly higher than from video alone. Studies from the Imperial College London show that surgeons trained on haptic simulators make fewer errors and achieve proficiency faster.

Remote collaboration. A team in a spatial computing environment shares presence: they perceive each other as occupying the same space, can gesture at shared 3D objects, and (with haptic integration) can physically manipulate shared prototypes. This is qualitatively different from a video grid of faces.

E-commerce. Feeling a fabric's texture, sensing a shoe's fit, testing a kitchen tool's weight and balance. Haptic interfaces convert "browsing" into "trying." The return rate for online clothing purchases (approximately 25-30%) is primarily driven by fit and feel issues that haptic interfaces could address.

Therapy and rehabilitation. Neural interfaces for patients with paralysis, haptic feedback for physical therapy, spatial computing for exposure therapy in PTSD treatment. The sensory internet's most immediate and unambiguous value is in medical applications where physical presence is therapeutic.