Last week the fine folks at USNI News and The War Zone reported on an intriguing interview with Rex Geveden. Geveden is the chief executive of BWX Technologies, the firm that supplies nuclear reactors to the U.S. Navy aircraft-carrier and submarine fleets. He forecast that the U.S. Navy’s next generation of nuclear-powered attack sub (SSN), currently dubbed SSN(X), “will be a larger type of submarine, probably in the size class of the Columbia”—Columbia being the new nuclear-powered ballistic-missile submarine (SSBN) class that rumbled into production this month when the navy awarded builders a $9.47 billion contract for the first hull and advance work on the second.

An SSN the dimensions of an SSBN? Doubtful. The Columbia class will reportedly tip the scales at over 23,000 tons, almost triple that of early-model Virginia-class SSNs now in service. (Later models have expanded in size and tonnage. Most notably, future Virginias will have a longer hull—adding more capacity to carry missiles or unmanned underwater vehicles, or UUVs. Length adds matériel adds heft. But even the bulked-up Virginia will be dwarfed by the Columbia.)

For comparison’s sake, the U.S. Navy’s Zumwalt-class destroyers, vessels of cruiser proportions, displace only about two-thirds as much as the Columbia SSBN. Or if you prefer your comparisons historical, the second Columbia will bear the name USS Wisconsin. The first Wisconsin, a battleship in Theodore Roosevelt’s Great White Fleet, displaced about half as much. Taking a Columbia on patrol will be like submerging a supersized battleship! In short, SSBNs are big vessels. They are built to crawl along slowly and silently beneath the waves, concealing their doomsday firepower underwater as a deterrent to nuclear attack. Attack boats are predators. As such they benefit from being nimble and fleet of foot.

So it’s hard to imagine an SSBN-sized SSN lumbering around in the coming decades. More likely Geveden was referring to the diameter of the new sub’s hull, which will apparently be around 40 feet compared to 43 for the Columbias and 34 for the Virginias. That’s in the Columbia range and, more to the point, is comparable to the three-hull Seawolf-class, which put to sea starting in the late 1990s. The Seawolf-class was built to duel a peer Soviet submarine fleet and remains at the forefront of undersea badassery. Navy leaders curtailed the class when it appeared the U.S. Navy would confront no rival of comparable stature any time soon. Shipwrights built the less pricey Virginia class for the low-threat world it appeared they would inhabit.

Back to the future. For the past couple of years navy magnates have made it plain that they want SSN(X) to be a “Seawolf-like submarine,” connoting Seawolf-like dimensions. That suggests a sub displacing over 9,000 tons when submerged—noticeably bigger than the Virginia boats but well under half the Columbia’s tonnage. You can foresee a boat like that sparring with China’s or Russia’s navy in the depths. There was less controversy to Geveden’s remarks than it first seemed.

Shift gears. Is SSN(X) the right boat to comprise the future submarine fleet? Let’s ask fleet designer extraordinaire Julian Corbett. Writing a century-plus ago, Corbett observed that age-of-sail fleets were trifurcated by function. The battle fleet, made up of capital ships capable of slugging it out with their peers in the line of battle, existed to destroy, scatter, or blockade an enemy battle fleet. Afterward ships of the line acted as the guardians of “cruisers,” smaller, more lightly gunned, and cheaper warships that were affordable in bulk. Cruisers fanned out to police the sea lanes once the hostile battle force was out of action; fleet units chased off or sank any remnants of the enemy fleet that contested friendly control of the sea. The “flotilla” was a collection of still smaller, more lightly armed (or sometimes unarmed), and inexpensive vessels that performed the administrative duties all navies must perform.

Corbett observed that naval technology had upended this neat division of labor, turning age-of-steam navalists’ world upside down. Torpedoes and sea mines—weapons small steam-driven ships of war could carry—had superempowered the cruiser contingent. Even the flotilla, heretofore an afterthought in sea warfare, had come to sport combat power. Commanders now had to fret about surface torpedo boats, submarines, and minelayers when bringing the battle fleet within reach of hostile seacoasts. Superempowered small craft could smite capital ships! Corbett bewailed the “historical revolution” that had rendered lessons from the past—meaning from the age of sail—suspect at best. Nor did his revolution ever subside. If anything it went into overdrive after his death with the debut of guided missiles, radar, nuclear propulsion, and other high-tech gadgetry for making war at sea.

Ships designed specifically to execute one of Corbett’s three functions are relics of the past. The functions themselves are not. In fact, they still provide a useful template for evaluating the design for any modern fleet—including the U.S. silent service. Nuclear-power advocates have dominated the submarine force since the days of Admiral Hyman Rickover, who remains a legend in the service. Echoing Rickover, they insist that the undersea fleet must remain an all-nuclear fleet. Each generation of attack boats will give way to the next into the indefinite future. In Corbettian terms, though, isn’t that akin to saying capital ships should constitute the whole fleet? If so, the submarine force appears destined to collide with the same challenges—affordability in particular, but also function—that kept age-of-steam navies from outfitting themselves entirely with battleships. No navy could afford such a fleet. Then as now, mixed fleets rode the waves.

An unaffordable subsurface fleet is a real prospect. Last year the Congressional Budget Office (CBO) forecast that the SSN(X) could cost $5.5 billion per copy. Navy estimates say $3.4 billion per hull—over 20 percent more than late models of the Virginia class, which run around $2.8 billion, but far short of twice the price. The price tag matters. The submarine force is set to dwindle to as few as 42 SSNs in the coming years as Cold War-era Los Angeles-class boats age out of their service lives. It will rebound as new Virginia-class construction replenishes the inventory. According to navy figures, the 355-ship fleet that the Trump administration has proposed and Congress endorsed would include 66 SSNs. Other plans bruited about peg the desirable number as high as 80 attack boats.

The more, the better. But it will be hard enough to sustain those totals at $3.4 billion per SSN(X)—nigh on unthinkable if the CBO’s $5.5 billion estimate pans out. A budget-busting fleet design is whimsy.

Fortunately, strategic overseers have time to think. The navy and shipbuilders are still defining parameters for the SSN(X) project. As Rex Geveden notes, the first boat will not take to the deeps until the late 2030s. If Corbett is waving a warning flag to alert navy magnates to the perils of an all-SSN fleet, his ideas about fleet design also hint at more plausible options for the U.S. submarine fleet as a whole. A contingent of Seawolf-like boats is eminently worth pursuing for open-sea combat, but other platforms may be able to do certain jobs better and more cost-effectively at certain places and times. To make the numbers work the silent service should seriously ponder fielding undersea analogues to Corbett’s cruisers and flotilla. They should follow time-honored practice.

Two candidates stand out. One will certainly constitute part of the future fleet, namely unmanned underwater vehicles of various types. Indeed, part of the rationale for a bigger SSN(X) is that it would be sufficiently capacious to operate UUVs of various types. This would be a hybrid arrangement in Corbett’s scheme of things. If unmanned technology pans out as envisioned, subsurface capital ships would act as motherships for elements of the flotilla while also bearing their own heavy armament to fight the good fight. Over time UUVs might—might—even qualify as cruisers, helping police the sea lanes in their own right. Like ships of the line watching over light sail-driven vessels in yesteryear, SSNs would act as protectors of uncrewed cruiser and flotilla contingents once the battle for maritime command was won.

All of that being said, the part UUVs will play in future operations remains somewhat hypothetical. It has to be subjected to the test of reality. Fleet experimentation and wargaming—rigorous and candid about the technology’s promise and limits—will be a must. Otherwise unpleasant surprises may befall the silent service in combat. That leaves manned but nonnuclear submarines as a proven option. Diesel-electric submarines will be part of any alliance war around the Eurasian periphery. For instance, Japan and South Korea operate impressive conventional submarines. Allied fleets fight as one. But there’s a case for diesel subs in the U.S. inventory as well.

Affordability is part of it. Take the Japan Maritime Self-Defense Force’s (JMSDF) latest state-of-the-art diesel boat, which is powered by lithium-ion batteries while underwater, as a standard of measurement. The Taigei will run $722.23 million according to official figures. That implies the U.S. Navy could acquire about four diesel boats for the price of one late-model Virginia. (CBO estimates the cost of a Virginia at $2.9 billion, slightly more than the navy tally and almost precisely four times the Taigei’s cost.) The ratio is even more lopsided when stacking the JMSDF boat against SSN(X). The navy could acquire as many as seven conventional boats for the price of one SSN(X), depending on whether the navy or the CBO estimate better approximates the true cost of the new SSN.