Greetings,

     I have a small company that works in the defense and telecommunications sector. In the coming months we have to train some of our engineers to develop on FPGA. The question is Xilinx or Altera?

  Which brand is the best to start? How is the industry standard?

Looking on the internet it seems that Xilinx has much more documentation and there is a more active community, but seeing companies like mine (at least in my country, Italy) everyone, absolutely everyone, uses Altera.

So my impression is that Xilinx has better support but most companies and products use Altera (Intel) FPGA, so at least in Italy, Altera is the industry standard. Is this my impression true?

Since I have to spend quite lot of money on training and software, I would like to make a conscious choice, Do you advise me to invest in training and software for Altera Intel FPGA or Xilinx FPGA?

Altera FPGAs are not intended for the military applications.

However, they are relatively cheap. For this reason, they are widely used in Russian armament.

On the contrary, the Xilinx FPGAs have special military lines of chips that are very reliable. They can be used even in aerospace devices.

Moreover, even the industrial Xilinx chips are rather reliable. This is proven by the respective reliability measurement datasheet. (I've designed the specific block with Xilinx FPGA operating at 130 Celsium degrees)

And finally, you can develop something using the industrial chips and then approach the military ones as needed. 

The second military FPGA company is Microsemi (former Actel).

It manufactures very reliable FPGAs with triple redundant circuits in SOI technology. They are used in aerospace devices very well.

Due to the radar tasks, my opinion is the following.

Xilinx FPGA architecture fits much better the high-speed DSP applications. The pipelined structures are mapped in these FPGAs very well. Due to this, the same blocks mapped into Xilinx FPGA have less hardware consumption (in LUTs or ALMs) and higher maximum clock frequency. The bit width of DSP48 multipliers fits the radar algorithms very well. The new FPGA series have a much higher volume of built-in RAMs which is very important for the radars. You can use the floating point data in Xilinx FPGAs as well as in Altera ones. The high-end Xilinx FPGA has built-in high-speed multichannel ADC, DAC so, it is intended to perform the 5-G communications and high-end radars. 

I concur, without a doubt, Xilinx is the leader in high performance and Military grade FPGA's, to reinforce the previous response