The war machine of the future is on its way. Unstoppable and invulnerable, it moves at terrifying speed, giving opponents no time to react. Enemies will not even see their human opponents, just a front line bristling with robots, unmanned aircraft and remote sensors.

This is the US army's future combat system (FCS), which will reach a critical decision point in April. If it is judged successful, within just five years the US will field a revolutionary new force.

The 60 ton tanks and 30 ton personnel carriers that rule the battlefield are on their way out. According to the new vision they are dinosaurs, too heavy to be flown to the scene of conflict and needing too much support. This new vision of the army is not "platform-centric", concentrating on vehicles, but "network- centric", built around a system of shared data.

The future soldier will not use heavy metal to protect him or strike the enemy; attack and defence will be a matter of data management. The system is a family of light ground vehicles with attendant robots and drone aircraft, seamlessly welded into a single fighting unit by a powerful and robust network. Just as an office PC network allows users to share data, the FCS network will allow troops to see and hear everything within range.

Sophisticated software will fuse the input from different sensors, building up a three-dimensional picture of the battlefield. Even the lowly infantryman will be wired, transformed into an objective force warrior with global positioning system (GPS), rifle-mounted video camera, laser rangefinder and data visor. "Network-centric warfare is to war what e-business is to business," says the Department of Defense's report to Congress.

"It has the potential to increase warfighting capabilities by orders of magnitude." The report explains how the new vision extends warfare from the physical into the information domain.

"The information domain is the domain where information lives. It is the domain where information is created, manipulated and shared."

Winning in the information domain means gaining information about the enemy while ensuring that the enemy is kept ignorant or deceived. The network-centric force can fight over a wider area with a smaller force, always one step ahead of the enemy, all of its units perfectly coordinated.

To realise this grand scheme requires hardware: the future combat system. The individual elements of the FCS will be smaller and lighter than existing armoured vehicles. The need to fit easily into Hercules transport aircraft limits them to about 16 tons. But the power of the network means they can together create a team stronger than the individual elements.

The army's precision fire is currently limited to line of sight, around five kilometres. The FCS will be able to attack targets 100 kilometres away using NetFires, known as "rockets in a box".

NetFires consists of a container with 15 missile launch tubes, which can be mounted on the back of an armoured vehicle or a truck, or left on the ground. When a commander identifies a target using imagery relayed from robot scouts, he can call up fire support at the click of a mouse.

NetFires identifies the nearest launcher and looses a missile, programming it with the target's details. The missile is equipped with a radar sensor and linked in to the network. It can loiter overhead for up to half an hour, so the target can be confirmed before it dives in for the kill.

Shorter-range weapons will include a new high-speed missile. This will have the same velocity as a tank shell, but without the weight and recoil of a tank gun. This fast missile will have precision guidance. Smart mortar and artillery shells will also feature, and for peacekeeping operations, the FCS will have a "scaleable lethality" weapon that can stun rather than kill.

The human soldiers will be kept at a safe distance from the front. The FCS vehicles will be hard to spot, with quiet engines and chameleon paint which changes colour to match the background.

Advocates say the FCS unit will see first and shoot first with deadly accuracy, denying the enemy the chance to get a shot in. However, there will still be a need for more protection than just lightweight armour.

This protection will come from the use of networked information. Attacks will be defeated by active measures. Radar, infrared and other sensors will detect incoming missiles and artillery rounds. They will then be engaged, by jamming or dazzling their sensors, or by firing projectiles to deflect or destroy them in a miniature version of Star Wars.

Smart armour will react by disrupting anything that strikes it. Lasers and high-powered microwaves are also being considered. Mobility and flexibility are part of the concept. The vehicles can be carried by helicopter or parachuted into action: a computer-controlled parafoil will land the FCS anywhere to within a few metres. The FCS-equipped force will require a fraction of the logistical support and backup of current forces. The light vehicles will have efficient engines, and ammunition will be a few smart rounds instead of thousands of dumb ones.

It will even be able to produce drinking water from the vehicle's exhaust. Robots are everywhere: small remote-controlled aircraft that look like flying buckets, and larger ones like the Predator drone. Land robots include portable ones for exploring buildings and autonomous armoured vehicles. The manned vehicles will be largely automated.

The FCS may even be able to drive itself. Some of the robots will be load carriers bringing supplies to the front line and transporting NetFires. The biggest challenge is developing a vehicle capable of navigating rough terrain, and a "follow the leader" approach is being evolved.

The lead vehicle will leave a trail of electronic markers for a convoy of robots, reducing the need for smart robots. The FCS is budgeted to spend $5 billion over the next five years. The lead contractor is Boeing, a company with no experience of building tanks but a long history in military electronics.

The FCS will be built around its electronic systems, as combat aircraft are built around their avionics (radar, control and communications), but costs will have to be much lower. Not everyone is happy with the network-centric vision. The US Army, 200 years old, values its traditions.

Soldiers are sceptical of newfangled hardware until it has been proven in action. In the 1991 Gulf war, US tanks shrugged off antitank weapons at point blank range and destroyed Iraqi tanks at will. Soldiers are concerned about trading their tanks for a "plastic tank" relying on untried technology for defence, and with little built-in firepower.

Some are horrified that FCS designs favour wheeled rather than tracked vehicles: they are better for long distances along roads, but inferior in cross-country ability. One critic describes it as a system that "cannot move, cannot fight and cannot survive if hit by enemy non-guided weaponry not subject to electronic gadget countermeasures, that relies on mouse-clicked pleas for bombardment from others to hopefully kill the enemy for it."

Rather than maximum technology, such critics want a force with fewer and more rugged gadgets. They point out that complex electronics can fail, and can be vulnerable. An opponent who brought down the FCS's networked communications could bring the force to a halt without firing a shot.

The robots would be paralysed, commanders left helplessly trying to reboot and NetFires silenced. The US army has had bad experiences with high tech. One was the MBT-70, a 1960s program to build a tank ahead of its time. This featured an advanced gun/missile launcher, special armour and a variety of features including a remote-controlled gun turret and a chemical toilet.

It was an impressive assemblage of technology, but was abandoned in 1971 after eight years of development, judged too expensive and too unreliable. The FCS program is ambitious, having been brought forward four years since the original plan. However, not all of its capabilities will be part of the first "Block 1" version. Block 1 vehicles will not have the full protection suite or chameleon paint, and the robotic elements will need human supervision. The biggest question mark hovers over the networking (see panel).

Existing secure military networks operate at speeds of 9.6 to 14.4 kbps; some doubt whether a 100 Mbps capacity, the sort that FCS is likely to require, will be ready before 2010. A decision to go ahead will be a bold one. Army budgets are tight - the billion-dollar Crusader artillery program was cancelled last year - and the FCS will eat up huge amounts of cash. The figure of $5 billion covers only development; production costs are anyone's guess.

The failure of any of the technology, including networking, sensors, robotics, protection or firepower could see the FCS joining the MBT-70. However, if the FCS performs as advertised, the US army's lead over the rest of the world's forces will become a gaping chasm.

Dominance of the information domain will be as important as dominance of airspace, with as much effect on ground troops. How will allies be able to fight alongside US forces if they cannot connect to their network? They might be a political necessity, but in the net-centric future, an ally outside the network will be as relevant on the battlefield as the regimental mascot.

Fitting the pieces together
The FCS data, voice and video communications will require major advances in networking technology. The program is called Mosaic - multifunctional on-the-move secure adaptive integrated communications. By 2004, it is expected to demonstrate a self-organising cluster of at least 15 nodes, with a data transmission rate between 56kbps and 15Mbps, with the lowest speed applying at the maximum separation of 100 kilometres. It will have to be secure, and highly robust, in the face of electromagnetic interference or jamming, while at the same time it must have a low chance of being detected by the enemy.

This probably means highly directional antennae operating across several different radio bandwidths. Reconnaissance units may rely on other means such as the laser communication used between stealth aircraft.

Bandwidth management is vital. The FCS will potentially have information flowing around it from multiple video feeds and other bandwidth-hungry inputs. It will only be able to handle these with a combination of powerful compression tools and intelligence built into the sensors so that they extract key information rather than transmit raw data.

There is a considerable potential for civilian spin-offs. A secure network that can be set up anywhere with a two-minute installation time and can link mobile users many kilometres away would solve many networking problems - if it is affordable.

· Comments to online.feedback @theguardian.com