Simple answer: they're cutting corners -- increasing shelf life, decreasing production costs, and overall increasing profits, like many of the big food corporations operating today.
Buying some filtered animal-derived blood plasma on the open market and letting the tissue grow/soak in it, would likely be a lot cheaper than precision mixing+dispersing of thickeners + reverse-pressure-gradient tissue impregnation of those thickeners. Food-grade blood plasma is the lowest-demand animal byproduct there is — it's what gets rejected out of even blood-sausage manufacture.
Same with collagen vs., specifically, carrageenan — collagen's cheap in bulk and works great for getting animal cells to stick to it; carrageenan's expensive, finicky to work with, and there are concerns about the carcinogenic effects of its long-term consumption. Many food-product manufacturers have moved away from previous formulations containing carrageenan; companies are only sticking with carrageenan at this point if there's nothing else that works within their constraints. Judging by other carrageenan-containing products, those constraints are probably something like "plant-derived; solid at room temperature; melts in your mouth; decent compressive strength, yet tears easily under tension."
And vegetable oils would be cheaper than animal fats... but vegetable oils with the same set of health guarantees as salmon (i.e. "omega-3 rich" vegetable oils) are not. And their product does claim to have the same health benefits as real salmon; so presumably they are aiming for that omega-3:omega-6 ratio target, since it's usually the headline "health benefit" of eating salmon. Which means they're probably buying, continuously-measuring, and mixing different oils to hit that ratio — similar to what orange-juice processors do to create a homogeneous juice.
While that type of infrastructure is certainly possible, there's no clear way to scale it like there is for cellular culturing.
And the result might be pretty disappointing even if you did it. Picture a human who spent their whole life in a coma in a hospital bed — they'd have zero muscle mass, fragile bones, etc.
However, you might be able to get around this by hooking the animal's body up to a synthetic brainstem that sends computer-controlled impulses down the spine – such that you've got e.g. a bunch of brainless salmon strapped in place in a flow tank, whose bodies are constantly being told to breathe, swim forward, open their mouths to capture food when food is released into the tank, etc.
I think that, besides the scalability challenges, the reason this approach isn't more looked into as an avenue of research, is that the ethics of cellular-cultured meat are really clear (there is nothing experiencing pain there; there aren't even nerve cells to sense or relay pain), while the ethics of brainless whole-animal cultivation are... non-obvious, to say the least.
> In addition to water and cell-cultivated salmon, our saku contains fats derived from canola, sunflower seeds, and algae, soy (an allergen), potato starch, konjac (a root vegetable), beta-carotene and lycopene (natural colors), carrageenan (an extract from red seaweed), and natural flavors.
Think about why each of these things are in there:
• Fats — because the parts [tissues] of the salmon that we eat, have not just muscle cells contained in them [the part that tastes + mouthfeels + cooks like salmon], but also fat cells (adipocytes), to contribute the taste + mouthfeel + cooking properties of "fatty tissue" [which is how we expect salmon to be] vs "lean tissue". And sure, the people creating this thing could have another tank growing "salmon-derived adipocytes", with some hormone bath to trick those adipocytes into absorbing and metabolizing nutrients from the environment to grow heavy with fat... but why bother? (That actually sounds dangerous, in fact — you might end up eating big doses of fish hormones trapped in the fat.) At the micro level, a little sphere of fat is a little sphere of fat; you can use a salmon adipocyte, some other kind of adipocyte, or even just a skin of sodium alginate, and the taste and texture of the result will be identical, as long as the fat inside the bag has identical properties (glyceride chain length, mostly).
• Natural colors and flavors — weirdly enough, because salmon by itself wouldn't look or taste fully like salmon; because the look and flavor come not just from what the salmon itself produces via the action of its cells/proteins/DNA, but also from "impurities" — things the salmon eats, that end up depositing into the salmon's tissues over time.
• Starch, caregeenan (and the implicit ingredient, water) — together, a simulacrum of (slightly-viscous) salmon blood. Using water alone wouldn't work; it's too thin, it'd just run out of the muscle tissue like a water from a sponge, desicating the tissue over a span of minutes. You need some thickener to prevent that. (I suppose you could make salmon blood plasma + platelets. Might be more nutritious if you did. Probably doesn't make much of a difference to taste or texture, but this might be a v2.0 goal of theirs.)
• Soy and konjac — a simulacrum of connective tissue, i.e. collagen. This is likely the matrix holding the cells in place. There's no such thing as "cells stacked directly on other cells"; there needs to be some non-cellular tissue matrix that binds them together. (Think "meat glue." Is a chicken nugget chicken?)
Why not actual collagen as a matrix, or maybe, say, gelatin? Why not ground-up shrimp as a colorant instead of beta carotent + lycopene? Why vegetable oils instead of animal fats? In all these cases, probably because their goal with these ingredients seems to be to only build this salmon out of plants + cells, rather than any animal byproducts. An unstated premise here seems to be that they want to design the process such that no matter how far it gets scaled up, there's no point at which it would be more economical to switch one of the ingredient sources from "make it in a bioreactor" to "get it from an animal byproduct sources", and at even further scale, "drive animal slaughter to get said byproduct as the product."
AFAICT, this is almost the closest thing you will ever be able to get to something you can call "salmon" — or maybe more specifically, "animal-harm-free salmon" — that can be created solely in a lab.
(To be wacky, you could, in theory, genetically engineer a... tree, or what-have-you, that would metabolically synthesize the salmon blood plasma, the salmon connective tissue, the salmon-prey-species tissue trace impurities, etc.; and also act as a host to a commensal salmon cell population; eventually putting all that together inside a fruit or something. Pluck and peel the fruit — salmon muscle tissue. We're probably 50 years from understanding genetic engineering well enough to do that; and even then, it'd probably be operationally impractical, due to salmon muscle tissue rotting at any temperature a tree would grow at.)