Pentagon impatience is the roadmap now: $61m for Smack, $500m for Scale AI

Smack Technologies, an Austin firm building battlefield AI, said on Monday that it had raised a $61m Series B, its second round of 2026, with military demand outpacing its ability to hire ^[1]. The more instructive number sits alongside it in The Next Web's account: Scale AI's recent $500m Pentagon contract, five times its previous deal, which implies a prior award in the region of $100m ^[2][3].

Read together, those figures describe a buyer rather than a market. Investor appetite follows the same signal, according to the report: a government that spent years apologising for its own slowness now wants software in soldiers' hands at start-up speed ^[4].

What that buyer is paying for is legible in Smack's product line. Omega is fires-planning software, the work of deciding what to strike, with what, and in what order ^[5]. It uses reinforcement-learning models and real-time computer vision to produce targeting options in minutes and to repair a disrupted battle plan in seconds, against staff processes that can currently take hours ^[6]. That is not a capability description so much as a stopwatch, and stopwatches are how this customer now writes requirements.

The second product is the tell. Alpha is a forearm-mounted display intended to replace the whiteboards and grease pencils still used to track a fight, and the company hopes to field between ten and twenty prototypes within six months ^[7]. Ten units is a small number. It is also a fielding commitment with a date attached, which is a different kind of promise from a demo.

The path to those contracts was partly handed to Smack. In March the Defense Department branded Anthropic a supply-chain risk and ejected it from classified work, which the report says prompted the Marine Corps and Navy to court smaller vendors ^[8]. By July, Smack had won two prototyping deals, one with the Joint Fires Network and one with the Marine Corps Warfighting Lab, together worth more than seven figures ^[9].

Hiring is the visible cost of being pulled forward. Smack went from nineteen employees in April to fifty-one, and expects roughly eighty-five by year end ^[10]: thirty-two net additions already, about 2.7 times its April size, with another thirty-four hires to go ^[11][12]. Chief executive and co-founder Andy Markoff argues the urgency is manufactured inside the Pentagon by planners who expect jammed communications and cut-off units, forcing decisions at the edge ^[13]. "The future of war," he said, "will be more decentralised than any conflict that we've ever fought" ^[14].

The unresolved part is oversight, and it is structural rather than rhetorical. Machine-speed targeting recommendations are exactly what ethicists and some lawmakers have warned about, because a faster loop leaves less room for meaningful human intervention ^[15]. Smack, like most of the sector, says a person stays in the loop; the sceptical reading is that "in the loop" degrades into approving outputs that arrive too quickly to interrogate ^[16]. The speed is the purchase.

Watch three things. Whether the ten to twenty Alpha prototypes are actually in hands inside six months ^[7], whether the July prototyping awards convert into anything larger ^[9], and whether "human in the loop" ever acquires a contractual definition rather than a press-statement one ^[16]. The sector context is not scarcity: tens of billions have gone into defence tech this year, with European buyers funding comparable work including Helsing's NATO-facing effort ^[17][18].