I'm going to split my answer into two parts.
1. Selective Breeding for Intelligence
It might be more helpful to estimate in terms of generations rather than years.
If cephalopods can produce a new generation every year (instead of a human generation's 15–30 years), that should reduce the amount of time a selective breeding program would require to select particular traits.
For comparison, consider fruit-fly breeding operations (which have a generation cycle of only ten days) and dog or fox breeding.
How long does it take to breed for a trait?
The Soviet scientist Dmitry Belyayev began a famous experiment in 1959 wherein he bred successive generations of wild silver foxes to produce tame ones, selecting based on their relative fear and friendliness toward humans with each generation. He was able to produce human-friendly foxes by the fourth generation and very friendly (eager for human contact) foxes after only six generations.
While repeated inbreeding can elevate the target trait, it can also exacerbate other traits unintentionally. For example, Belyayev's foxes started getting multicolored coats after 8–10 generations. After 15–20 generations some of the foxes developed abnormalities like shorter legs and underbites or overbites.
That's just friendliness. What about other traits?
Breeding for friendliness could certainly get you to an offshoot that is more collective and cooperative, but that's a far cry from producing a species capable of abstract thought.
It worked well for Belyayev because the trait they were shooting for also made the animals easier to handle and thus easier to breed. In contrast, his counter experiments breeding animals for greater fierceness had to be discontinued after fewer generations due to difficulty handling the animals and keeping them from killing each other.
To selectively breed for some other trait besides friendliness you need to have a way to identify the trait in the available breeding population. If your organization has access to modern (or even 70s-era) technology, they may be able to identify the genes involved in learning and memory, as has been accomplished with fruit flies, so that they can more quickly identify which subjects to include or exclude from the program. Furthermore, artificial insemination allows them to explore behavioral avenues that might otherwise result in non-breeding.
Both of these (genetic testing and artificial insemination) would make it faster to isolate and enhance aspects we associate with intelligence than such aspects could develop by natural evolution.
But that's assuming the aspects are already available or incipient in the genes in the first place; you can't use selective breeding to produce a fruit fly with lungs, for example.
To produce intelligence ex nihilo could take millions of generations, simulating natural evolution.
Cephalopods to the rescue
Fortunately, you picked cephalopods as your seed creature, so you don't have to produce intelligence ex nihilo!
There's some evidence that at least some species of cephalopods are already capable of abstract thought, including problem solving and tool use. Some cephalopods (specifically octopus and cuttlefish) have the highest brain-to-body mass ratio of all invertebrates. Cephalopods have been observed opening screw-capped jars, throwing rocks to smash aquarium glass, and stealing food from lobster traps and even from boats.
Some of your climate-protection organization's targets will be more difficult to reach than others. In particular, the ability to pass knowledge on to the next generation will probably require some kind of language or mathematical system. The ability to produce and understand language may be difficult to isolate and exaggerate in the cephalopod genome, since their most advanced communication at this point seems to be flashing warning signs and mating displays with the pigmentation on their skin.
I imagine the organization would be able to produce different levels of intelligence over the course of many generations; the end goal of a communicative, language and math-savvy species may require millions of years, but in the shorter term they may develop something intelligent and crafty, maybe even individually mathematically-savvy, but lacking the ability to communicate complex thoughts to others of its kind.
2. Breeding Cephalopods in Captivity
I'm going to assume the chosen species for the breeding program is a type of octopus for the remainder of this answer. They have a large brain to body mass ratio and have been kept in captivity both in professional and personal aquariums.
Guides for keeping pet octopuses recommend tank sizes in excess of 50 gallons.
This pop-science article from 2015, Does an Octopus Make a Good Pet, states the following:
the animals need at least a 55-gallon aquarium with a second large tank to hold filtration equipment.
This Keeping Cephalopods guide published in Marine World Magazine in 2004 has a more modest recommendation:
The minimum size should be at least 36×18×18 (inches) to be used for small octopus species and as big as you can get after that.
36 × 18 × 18 inches = 11664 cubic inches = about 50.5 gallons
Those size recommendations are for single octopuses, under the assumption that they will be kept solitary to prevent cannibalization. If the breeding program succeeds in reducing fierceness, producing more social/friendly critters, cannibalism might be less of a concern after a few generations, in which case multiple octopuses could be kept in the same tanks.
Octopuses are more sensitive to temperature, salinity, and ph balance than typical aquarium fish, so you should assume that the tanks will require significant ongoing maintenance and filtration.
Octopuses are predators and require a regular diet of crustaceans and/or mollusks (preferably live), so the organization would either need some advanced surreptitious aquaculture or a reliable external supply chain by which to obtain feeder animals. This might be easier if their operation is situated near an ocean.
An octopus only lives 6 months to 2 years. To maximize the opportunity for the captive critters to produce viable offspring, it'd behoove the organization to seek ways to reduce the mortality rate.
Normally when an octopus breeds it dies shortly thereafter (the male after spawning, the female after brooding), but research suggests this can be prevented by removing their optic glands, implying that glandular secretions directly trigger the senescence that leads to their mortality. You could conceivably prevent this effect and prolong their longevity through surgery or selective breeding.