Smart Cities
How AI-Prioritized Traffic Signals Are Cutting Transit Delays in San Jose
Published by IOT San Jose Research & Editorial Team
- What Transit Signal Prioritization Actually Does
- Why San Jose Is a Useful Case Study for Other Cities
- Implementation Lessons for Local Governments and Enterprises
- The Business Value Beyond Transit
- Common Challenges Cities and Enterprises Run Into
- What Comes Next for Connected Traffic Systems
- What This Means for Budget Planning
- How the Priority Decision Actually Gets Made
San Jose's buses are not getting faster because the city bought new buses. They are getting faster because traffic signals now recognize a bus is coming and hold the green light for it. That single change, rolled out across 24 routes in early 2026, cut average bus travel time by 20 percent and slashed red light wait times by half during an earlier pilot phase. It is a small technical adjustment with an outsized result, and it is a useful window into what a mature Internet of Things deployment actually looks like once the pilot stage ends.
Most articles about smart traffic still describe the idea in the abstract: sensors talk to signals, signals adjust in real time, congestion improves. San Jose's rollout is worth examining precisely because it moved past that abstraction. The city, working with a company called LYT and the Valley Transportation Authority, built a system that makes thousands of individual signal decisions a day based on live vehicle location data, and it did so without ripping out existing signal hardware. That distinction, between replacing infrastructure and layering intelligence on top of it, is the real story for any city or business evaluating a similar investment.
What Transit Signal Prioritization Actually Does
Transit signal prioritization, often shortened to TSP, is not a new concept. Cities have experimented with bus priority signals for two decades using basic radio beacons that request a green light when a bus approaches an intersection. What changed in San Jose is the decision layer sitting behind that request. Instead of a fixed rule that always favors the bus, the system ingests live GPS data from the transit fleet, cross references it against current signal timing and downstream congestion, and calculates whether extending a green phase actually helps the bus without creating a worse backup for cross traffic.
That distinction matters because naive priority systems have a bad reputation among traffic engineers. Give every bus an automatic green light and you eventually create gridlock on the perpendicular street, which then delays emergency vehicles and everyone else. San Jose's deployment avoids that failure mode by making the priority decision conditional and continuously recalculated, not a static override.
The Numbers Behind the Rollout
- A two route pilot launched in 2023 reduced red light wait times for buses by 50 percent.
- The citywide expansion to 24 routes in 2026 produced a 20 percent increase in average bus speed.
- The technology partner, LYT, has been developing the underlying platform for roughly a decade and has received National Science Foundation grants focused on reducing congestion and emissions.
- City officials reported that buses are now sometimes running ahead of schedule, a problem transit planners rarely had to solve before.
Why San Jose Is a Useful Case Study for Other Cities
San Jose did not build this system on a blank slate. Like most mid-size American cities, it has a mix of signal controller hardware from different eras, some newer adaptive controllers and plenty of older fixed-time units. A common assumption among city IT leaders is that a smart signal upgrade requires replacing that hardware first. San Jose's approach shows a cheaper path: add a connected decision layer that communicates with existing controllers through standard interfaces, and reserve the expensive hardware replacement for locations where it is genuinely necessary.
This is where the business case becomes relevant beyond transit agencies. Any organization managing a distributed fleet, delivery vans, service trucks, shuttle buses for a corporate campus, faces the same question: does real-time coordination require new vehicles and new infrastructure, or can existing assets be connected through a software and sensor layer instead. San Jose's experience suggests the second path is both faster and cheaper to prove out.
Implementation Lessons for Local Governments and Enterprises
Organizations considering a similar deployment, whether a city planning department or a private employer managing a shuttle fleet across a Silicon Valley campus, tend to run into a consistent set of decisions.
- Start with a bounded pilot. San Jose tested on two routes before expanding to 24. That gave engineers real data on edge cases, like what happens when two priority requests arrive at the same intersection within seconds of each other, before scaling.
- Treat data quality as the actual product. The system depends on accurate, low latency GPS data from every vehicle. Any gaps in that data stream degrade the accuracy of every signal decision downstream.
- Coordinate with signal ownership early. Traffic signals in most California cities are owned and maintained by the public works or transportation department, a separate entity from transit agencies. San Jose's project required tight coordination between VTA and the city's Department of Transportation from day one.
- Measure the right metric. Average speed improvement is useful, but on-time performance and reliability, not just raw speed, are what riders and dispatchers actually notice.
The Business Value Beyond Transit
It is tempting to file this under public sector innovation and move on, but the underlying pattern applies directly to private sector logistics and facilities operations in the South Bay. Companies operating corporate shuttle networks between San Jose, Sunnyvale, and Mountain View face the identical problem transit agencies face: fixed signal timing that ignores real vehicle position, leading to unpredictable arrival times and wasted driver hours idling at red lights. The same connected prioritization logic, scaled down, applies to private shuttle fleets, last-mile delivery vehicles, and even emergency response vehicles operated by large campuses and hospital systems.
There is also a data dividend that is easy to overlook. Once a city instruments its signal network for real-time vehicle prioritization, the same sensor and data pipeline becomes a live congestion dataset that planning departments, logistics companies, and even retailers can use for demand forecasting and route optimization.
Common Challenges Cities and Enterprises Run Into
| Challenge | Why It Happens | How San Jose Addressed It |
|---|---|---|
| Conflicting priority requests | Multiple buses converge on the same intersection during peak hours | Dynamic recalculation instead of fixed override rules |
| Legacy signal hardware | Not every intersection has modern adaptive controllers | Layered a connected decision system on top of existing controllers |
| Stakeholder coordination | Signals and transit are usually managed by separate agencies | Direct partnership between VTA and city transportation staff |
| Public skepticism | Past smart city pilots have overpromised results | Published measurable before and after data from the 2023 pilot before scaling |
What Comes Next for Connected Traffic Systems
San Jose officials have already signaled interest in extending similar prioritization logic to emergency vehicles, which raises a harder coordination problem than buses because fire trucks and ambulances do not follow fixed routes. Expect the next phase of this technology to focus on dynamic, on-demand priority requests rather than route-based ones. Freight corridors are another likely extension, particularly around the logistics hubs along Highway 101 and Interstate 880, where delivery reliability has direct cost implications for regional distribution companies.
The broader trend is a shift away from treating traffic signals as static infrastructure and toward treating them as one more connected endpoint in a live operational network, alongside enterprise IoT deployments and municipal systems for smart city management. For a region built on the assumption that infrastructure should be programmable, that shift was probably inevitable. The question now is how quickly other California cities follow San Jose's model rather than starting their own pilots from zero.
What This Means for Budget Planning
One reason this project deserves attention from city budget officers, not just transportation planners, is what it did not cost. Full adaptive signal replacement programs typically run into the tens of millions of dollars for a mid-size city and take years to permit and construct, since each intersection often requires new controller cabinets, conduit, and utility coordination. San Jose's approach avoided the bulk of that expense by working through existing controller interfaces and a cloud-based decision layer, which is the same reason the two-route pilot in 2023 was able to expand to 24 routes within roughly two years rather than a full decade-long capital program. For finance departments evaluating IoT investments broadly, that gap between infrastructure replacement cost and software-layer deployment cost is often the single biggest factor in whether a pilot ever turns into a citywide rollout.
That cost structure also changes who can reasonably attempt a similar project. A software and sensor layer that rides on top of existing signal hardware is within reach for mid-size suburbs and smaller transit agencies that could never justify a full adaptive signal replacement program. Expect vendors in this space to increasingly pitch the technology on that basis, positioning it as an accessible middle step between doing nothing and undertaking a full infrastructure overhaul.
How the Priority Decision Actually Gets Made
It helps to walk through what happens in the seconds before a bus reaches an intersection. The vehicle's GPS position streams continuously to the platform, which calculates the bus's estimated arrival time at the upcoming signal. That estimate is checked against the signal's current phase and the state of any other priority requests nearby. If extending the green phase by a few seconds would let the bus through without creating an unacceptable delay for the conflicting direction, the system sends a request to the local controller to hold or advance the phase. If the calculation shows that granting priority would create a worse outcome elsewhere, no request is sent, and the bus waits through a normal cycle like any other vehicle. That conditional logic, recalculated at every intersection in real time rather than applied as a blanket rule, is what separates a genuinely adaptive deployment from the simpler priority beacon systems cities experimented with in the 2000s and 2010s.
Frequently asked questions
Related articles
Smart Cities
Smart Cities
Smart Streetlights and Traffic Sensors: San Jose's Approach to Silicon Valley Gridlock
Smart Cities
San Jose's Digital Inclusion Fund: Bridging the Broadband Gap With IoT Infrastructure
Smart Cities
