The Satellite Is Not the Point: Loft Orbital’s Paul Lasserre on Autonomous Intelligence, African Sovereignty, and the Code That Will Define Who Controls the Sky
There is a provocation embedded in the way Paul Lasserre talks about satellites, and it is entirely deliberate. The General Manager for AI for Space at Loft Orbital does not speak about his company’s technology the way most people in this industry speak about hardware. He does not lead with sensor specifications or orbital parameters. He leads with a question: What is the job to be done? And his answer to that question, applied specifically to Africa, is reshaping how a growing number of agencies on the continent think about what they actually need from space.
Lasserre arrived at the NewSpace Africa Conference 2026 in Libreville representing a company that has spent years repositioning what a satellite is supposed to be. Loft Orbital, the San Francisco-based satellite infrastructure company that secured EUR 170 million in Series C funding in January 2025, bringing its total capital to over EUR 300 million, is not building cameras in orbit. It is building, as Lasserre puts it plainly, intelligent autonomous systems that happen to operate from space. The distinction matters more than it might initially seem, and for African governments, agencies, and startups, it carries implications that go well beyond the technical.
From Hours to Minutes: The Time-to-Insight Problem
To understand what Loft Orbital is offering, it helps to understand the problem it is solving. In a conventional Earth observation architecture, the chain from event to decision runs through several stages: a satellite acquires an image, downlinks it to a ground station, the raw data is processed, and the resulting information is disseminated to those who need it. In the best case, that chain takes hours. Often, it takes the better part of a day.
For a disaster response team managing a flood in a low-lying river delta, or a conservation authority trying to locate the origin point of a wildfire in a remote national park, a day is not a timeline. It is a catastrophe.
Lasserre describes the changes that occur when a trained AI model runs directly on the satellite. “Once the satellite flies over the event, detection takes roughly one second,” he says. The alert is then downlinked not as a raw image tile requiring ground processing, but as a structured event, a precise location, a timestamp, and a risk score. The time-to-insight collapses from hours to minutes. The bottleneck in the system is no longer the space segment. It is the ground distribution channel, which is a far simpler problem to solve.
He is specific about the use cases where this matters most. Not the visible disasters that humans are already watching, but the invisible ones. A methane leak on a pipeline that nobody inspects weekly. An oil spill far offshore, beyond the reach of coastal patrols. A wildfire began in a remote park before anyone had reported smoke. “These are exactly the situations where nobody is watching,” he says, “and where an autonomous satellite that knows what to look for becomes the early warning system you did not have before.”
An Open Platform, Not a Closed Stack
The natural follow-on question for African developers and startups is whether this infrastructure is accessible to them or is another layer of technology designed in San Francisco and consumed on the continent, with African hands never touching the code.
Lasserre’s answer is simple. “We are building a platform, not a closed stack.” The vision for Loft Orbital’s application marketplace is explicitly modelled on the logic of modern developer infrastructure: self-service onboarding, standardised procurement, billing and metering tools, and a pathway for any organisation with a credible use case and real end customers to deploy its algorithms on Loft’s platform.
The constellation, built around the Altair programme via Orbitworks, the joint venture between Abu Dhabi-based Marlan Space and Loft Orbital, is launching this year with commissioning expected in 2027. At this early stage, the platform is working with a small number of launch partners, a deliberate quality-over-quantity decision while the system matures. But the direction of travel is clear.
“African-built code on the Loft platform is not a future ambition,” Lasserre says. “It is exactly the developer journey we are designing the marketplace around.” He is direct about the constraint today: the maturity of the product platform, not the willingness to onboard African talent. Startups from Kigali’s innovation ecosystem, Lagos (or anywhere else on the continent), with a credible use case and identifiable end customers are, in his words, welcome in the early cohort.
Sovereignty Without the Wait
One of the more structurally important arguments Lasserre makes concerns how African nations should think about space sovereignty in an era when the options available to them have changed fundamentally. The traditional framing presents a binary: a nation either owns and operates a fully bespoke national satellite, or it purchases commercial data as an external customer with no real control over the infrastructure. He considers that binary obsolete.
Altair, the ten-satellite AI-enabled Earth observation constellation being integrated by Orbitworks at its 50,000-square-foot facility in Abu Dhabi’s KEZAD economic zone, is built on Loft Orbital’s Longbow platform, a derivative of the flight-proven Airbus Arrow bus, featuring sub-metre optical, shortwave infrared, thermal, hyperspectral, and radio frequency sensors, paired with heavy onboard computing for real-time in-orbit data processing. Launches are scheduled to begin in the second half of 2026.
What Altair makes possible for an African nation, Lasserre argues, is access to that capability stack from day one, without the five-year build timeline and the single-generation lock-in of a bespoke national programme. He deliberately reaches for the cloud analogy. “Nobody in 2026 builds a data centre before launching an application,” he says. “You build on AWS or Azure, you keep control of your data and your intelligence layer, and if sovereignty demands it later, you move specific workloads in-house.”
His view on which layer of the stack actually constitutes sovereignty is pointed and worth sitting with. “The layer that matters is not the hardware,” he says. “It is the mission, the models, and the decision layer. Those are sovereign by nature, and African states should own, operate, and protect them.” The implication is that an African agency that develops and controls its own models and decision systems has more meaningful sovereignty than one that owns an ageing satellite but relies on foreign partners to interpret its data.
The Economics of Intelligence Over Bandwidth
A persistent structural constraint for African space users has been the high cost and limited bandwidth of ground station infrastructure. Lasserre frames this not as a problem to be solved through infrastructure investment, but as one that AI largely eliminates.
The reduction in downlink volume his system achieves is, by his account, more dramatic than the 80-90% figure sometimes cited in industry discussions. “Once you process data on the spacecraft, the reduction is above 99% for the real-time layer,” he says. The comparison clarifies the logic: a conventional operator generating and downlinking terabytes of raw imagery per day incurs a massive bandwidth cost for data that, in most cases, contains nothing of interest. With onboard AI, what is transmitted in real time is a structured detection event, a position, a timestamp, a risk score, and perhaps a small compressed image. That is hundreds of kilobytes, not terabytes.
Two structural consequences follow. First, an agency downlinks only what it actually wants to examine in detail, rather than blindly dumping everything. Second, because the system is no longer constrained by downlink capacity, sensors can remain effectively open across far larger areas, more land, more ocean, more coastline, more border, without a proportional increase in cost. The option to downlink the full high-resolution scene for ground analysis remains available, but the bandwidth cost is incurred only when justified. “For an African agency,” Lasserre says, “this inverts the economics. You stop paying to downlink what you will never look at. You start paying only for intelligence.”
On Liability, Governance, and Honest Probabilism
The deployment of autonomous AI systems in high-stakes environments inevitably raises governance questions, and Lasserre is candid rather than evasive about them. The right proxy, he argues, is how probabilistic AI is already handled in comparably regulated industries. Medicine uses AI-assisted diagnostics. Finance uses algorithmic risk systems. The space industry is not creating a new legal category; it is joining a conversation that other sectors have already been having.
The governance model he describes is a layered responsibility framework. The platform is one layer. The model is another. The training data is a third. Crucially, the model and training data come from the partner or end customer, not from Loft itself, which means accountability is distributed across the chain rather than concentrated in the infrastructure provider.
He insists on honesty about the system’s limitations. False negatives are a physical reality: clouds, revisit gaps, and sensor geometry mean that no orbital system catches everything, and claiming otherwise would be a disservice to operators. False positives are equally inevitable in any probabilistic system. The answer, he argues, is not to pretend otherwise, but to design operational workflows around probabilities rather than certainties.
His example is instructive. A maritime patrol aircraft that previously conducted random inspections of vessels now receives a ranked list of targets, each with risk scores for oil spilling, suspicious rendezvous, or dark-vessel activity. Even if some alerts are false positives, the aircraft’s operational effectiveness improves dramatically compared to random patrolling or no monitoring at all. The same logic applies to wildfire detection, border surveillance, or fisheries enforcement. A probabilistic alert with a known false-positive rate is vastly more useful than no monitoring, or than monitoring only the small fraction of territory that can be covered by conventional means. “The AI assists the decision maker,” he says. “It does not replace them.”
The Code That Controls the Sky
Two of the most direct exchanges in Lasserre’s conversations at Libreville concerned the long-term trajectory of African space ambitions and his willingness to state plainly what he believes about them. Speaking on the Kratos Constellations podcast, he had noted that AI is quietly reshaping who directs outcomes in orbit. At a conference with more than 20 African space agencies in attendance, the question of what that means for their futures is not abstract.
His answer is consistent and unapologetic. “The strategic advantage in space over the next decade will not be measured in kilograms to orbit,” he says. “It will be measured in code deployed to orbit.” Rockets matter. Satellite buses matter. But they are, increasingly, becoming commodities. Value capture has moved up the stack to the model, application, and decision layers.
For African agencies, the practical implication is specific. They do not need to wait for a national launcher or a national satellite factory to begin building sovereign intelligence capability. They can start today by developing applications on top of the infrastructure being brought to orbit, training analysts on real orbital data, and building the domestic ecosystem that will make truly sovereign assets viable later. “The agencies that move first will shape their national doctrine on autonomous space systems,” he says.
He is careful to frame this not as a replacement for national space ambitions but as the fastest pathway to making those ambitions real. Some use cases may eventually justify sovereign-owned assets. When that moment arrives, the argument for building them will be far stronger if the country has already demonstrated what it can do with the data. But starting from hardware, he argues, means spending years and substantial budgets on infrastructure that will often already be a generation behind by the time it reaches orbit. “Start from the job to be done,” he says. “Identify the outcome, leverage the infrastructure we are launching this year, work backwards.”
The continent he is describing, one with vast territories and limited ground-based assets, early warning needs across fisheries, borders, pipelines, and forests, strong software talent, and shorter legacy commitments to obsolete industrial models, is not incidentally well-suited to Loft Orbital’s offer. It is, in his assessment, the environment where the leapfrog is most structurally possible.
Why Loft Orbital Came to NSAC
Lasserre’s description of why Loft Orbital attends the NewSpace Africa Conference is operational rather than diplomatic. Africa, he says, is structurally well matched to what Loft and Orbitworks offer: fast innovation cycles, strong cloud and software talent, vast territories where human or drone surveillance is unrealistic, and a geopolitical environment where early warning from space serves as a genuine form of national insurance. The hypothesis driving Loft’s engagement is that countries with less legacy space heritage are best positioned to leapfrog precisely because they are not locked into the industrial assumptions of the previous generation.
He measures the impact of conference participation in one currency: concrete conversations that convert to pilots or MoUs within six to twelve months. Not badge counts, not panel appearances.
His suggestion for future editions of the conference reflects the same impatience with narrative over execution that runs through everything he says. He would like to see working sessions alongside the panels, structured workshops where African agencies and potential partners co-design use cases, map ecosystems, and identify which countries have the political will and the anchor users to move first. “That is how transformations actually move from conference to reality,” he says. “Not through declarations, but through rooms where people leave with concrete next steps.”
His closing message to African agencies is, characteristically, stripped of embellishment. Early warning from space is not an abstraction. It is a strategic capability, and the countries that invest in it, whether for illegal fishing, border monitoring, pipeline surveillance, deforestation, or fire detection, will be the ones that protect their populations and their resources most effectively. The countries that do not will continue to be informed by others about events on their own territory. “Africa has a once-in-a-generation opportunity to build that capability on its own terms,” he says. “Not by replicating the industrial space model of the twentieth century, but by owning the layer that actually creates value: the models, the data, the decisions. We are here as a platform, not as a vendor. The invitation is open.”