Berkshire Q1 2026: $1.72B Insurance Gain as Abel Maps GEICO's Five-Year Tech Rebuild

From analyzing GEICO's rate filings across 20 states over the past year, the 87.3% combined ratio masks a deliberate trade: Abel is sacrificing premium growth to fund a tech rebuild that will not pay off for three to five years. Written premiums grew just 1.5% in Q1 2026 while Progressive's book expanded by 9% in policies in force. That is not a competitive failure. It is the signature of a carrier that decided to fix its data infrastructure before reopening the growth throttle, and Abel said as much at the Omaha meeting: "It's not going to be easy to just restart the growth engine." The underwriting profit finances the rebuild. The rebuild is supposed to produce the next decade of growth. Whether that sequence works depends on execution that has not yet been demonstrated.

$1.72B

Berkshire insurance underwriting profit, Q1 2026 (up 29% YoY)

87.3%

GEICO combined ratio, Q1 2026

$397B

Berkshire cash and Treasury holdings, a record

The Q1 2026 Insurance Results: Segment by Segment

Berkshire's total operating earnings hit $11.35 billion in Q1 2026, up 18% from $9.64 billion a year earlier. Insurance accounted for the largest share: $1.72 billion in underwriting profit plus $2.68 billion in investment income on the float. The underwriting gain alone represented a 29% year-over-year increase, driven primarily by the absence of Q1 2025's California wildfire charges in the reinsurance segment.

The segment-level breakdown shows where the profit came from and how much GEICO dominates the underwriting picture.

Insurance Segment	Q1 2026 UW Profit	Q1 2025 UW Profit	Change
GEICO	$1.416B	$2.173B	-34.8%
BH Primary Group	$476M	$(144M)	Turnaround
BH Reinsurance Group (P&C)	$637M	$68M	+837%
Life & Health Reinsurance	$126M	$70M	+80%
Total Insurance Underwriting	$1.72B*	$1.34B*	+29%

*Consolidated figures after intercompany eliminations. GEICO's standalone $1.416B profit is the largest single-segment contributor but the total reflects netting across segments.

The GEICO result deserves careful reading. The $1.416 billion underwriting profit is strong in absolute terms but represents a 34.8% decline from Q1 2025's $2.173 billion. That year-over-year drop has three visible drivers. First, losses and loss adjustment expenses increased $853 million, reflecting higher claim frequencies and average severities across private passenger auto property damage, collision, and bodily injury coverages. The favorable frequency trends that had persisted through 2024 reversed in early 2026. Second, underwriting expenses jumped $338 million (29.3%), driven by what the 10-Q describes as a "steep increase in policy acquisition expenses tied to advertising." Third, written premiums grew only $168 million (1.5%), providing minimal top-line offset to the cost increases.

The reinsurance group's turnaround is the other major story. Q1 2025 included roughly $860 million in catastrophe charges from the California wildfires. With no comparable event in Q1 2026, P&C reinsurance swung from $68 million to $637 million in underwriting profit. The BH Primary Group's reversal from a $144 million loss to a $476 million gain reflects loss and LAE declines of $660 million, partly related to more favorable prior-year development.

GEICO's 87.3% Combined Ratio: Decomposing the Walk

The 87.3% combined ratio is exceptional by any industry standard. For context, AM Best's 2026 P&C industry composite projects a 96.9% combined ratio. GEICO is running nearly 10 points better. But the components tell a more nuanced story than the headline suggests.

Component	Q1 2026	Q1 2025 (Approx.)	Walk
Loss ratio	73.9%	~69.0%	+4.9 points
Expense ratio	13.4%	~10.8%	+2.6 points
Combined ratio	87.3%	~79.8%	+7.5 points

A 7.5-point deterioration in a single year is material, even when the starting point was an unsustainably low 79.8%. The loss ratio increase to 73.9% reflects the reversal of favorable claim frequency trends that had been GEICO's tailwind through most of 2024 and 2025. Both frequency and severity are now running against the book: higher claim counts per policy and higher average cost per claim across liability and physical damage coverages.

The expense ratio climb to 13.4% is more telling. GEICO's expense ratio has historically been among the lowest in personal auto, a structural advantage of the direct model that avoids agent commissions. The 2.6-point increase signals that Abel is spending on customer acquisition at a rate that exceeds the premium growth it generates. Advertising expense is the largest driver. This is consistent with the "unprecedented shopping activity across the auto space" that Abel described at the annual meeting, but it also means GEICO is paying more per new policy at a time when retention is under pressure from Progressive's telematics-driven pricing precision.

Why 87.3% Is Still Excellent Despite the Deterioration

An 87.3% combined ratio means GEICO retained 12.7 cents of underwriting profit on every dollar of earned premium before investment income. Compare that with the personal auto industry average that typically runs between 98% and 102%. The deterioration from 79.8% is not a sign of distress; it is a normalization from what was likely the most profitable quarter in GEICO's history. The 79.8% included exceptionally favorable frequency, minimal catastrophe exposure, and a low advertising spend that was suppressing new business acquisition. An 87% to 90% range is closer to GEICO's long-run sustainable operating point, and it still generates the cash flow to fund Abel's technology investment thesis.

The Growth Gap: GEICO's 1.5% vs. Progressive's 9%

The starkest contrast in Q1 2026 personal auto results is the growth differential. GEICO's written premiums grew $168 million, or 1.5%, reaching $11.7 billion for the quarter. Policies in force grew just 2% year over year. Progressive, by comparison, grew policies in force by 9% (with personal auto alone up roughly 11%), added approximately one million policies in Q1 alone, and reported net premiums earned of $20.97 billion.

Metric	GEICO Q1 2026	Progressive Q1 2026
Written premiums	$11.7B	$23.6B
Written premium growth	1.5%	~6.5%
Policies in force growth	2%	9%
Combined ratio	87.3%	86.4%
UW profit trend	Down 34.8% YoY	Net income up 9.8% YoY

This is not a story about one carrier outperforming another in a single quarter. It is the visible result of strategic choices made years earlier. Progressive invested in telematics starting in 2010 and now has 21 million connected policyholders generating real-time driving data that feeds a pricing model no competitor can replicate without equivalent time invested. GEICO, under Buffett's capital-light philosophy, spent the 2010s as an advertising-driven direct writer that competed on brand awareness and price rather than on data granularity.

The result is visible in the rate filing patterns. Progressive's filings reflect telematics-segmented rates that allow it to grow aggressively in preferred segments while maintaining margin. GEICO's filings are broader, less granular, and increasingly reliant on blunt frequency and severity adjustments rather than individual risk-level pricing precision. Abel inherited this gap when he took the CEO role on January 1, 2026, and the technology rebuild is his response.

Abel's Three "Narrow AI" Principles

At the May 2-3 annual meeting in Omaha, Abel laid out his AI philosophy in terms that diverge from the language used by most carrier CEOs this earnings season. Where AIG's Peter Zaffino talks about agentic multi-model orchestration and Chubb's Evan Greenberg describes 85% automation targets with 1.5 combined ratio points in savings, Abel used the phrase "narrow AI" and outlined three principles for how Berkshire will deploy it.

Principle 1: Human governance over AI recommendations. "There's human involvement. Our managers, our employees are involved," Abel said. This is not just boilerplate. Abel described a framework where AI augments underwriter capacity rather than replacing underwriter judgment. The specific example he gave was underwriter coverage: "If we were looking at a risk, and we had our traditional underwriters doing it, we might have looked at the five largest risks. Now, we can pretty much in a fairly quick way, yes, focus on those, but get a view on another 15 risks using technology." Ajit Jain confirmed this characterization as "an accurate assessment of the evolving use of technology within the insurance businesses."

Principle 2: Repeatability safeguards on AI outputs. Abel described a specific validation methodology: "If we ask it to ignore today's information and just focus on yesterday, and we get the same answer, then we've safeguarded the system." This temporal consistency test is a form of output stability validation that ensures the AI produces reproducible results rather than drifting with each input change. It aligns with the model validation standards actuaries are familiar with under ASOP No. 56 (Modeling) and the NAIC's emerging AI governance framework.

Principle 3: Value-focused deployment. "We're not going to do AI for the sake of AI," Abel stated. "It has to be additive to our businesses." And: "We will deploy AI on a small scale, focusing on ways that create value." This contrasts with the Travelers approach of deploying 10,000 Claude assistants across the enterprise or the AIG model of building multi-agent orchestration systems. Abel is describing incremental, ROI-justified deployments rather than platform-level transformation.

Jain reinforced the conservative stance: "Right now, what we are seeing is AI being used as a productivity tool, as a mechanism for reducing labor costs and doing routine, repetitive things." He went further on limitations: "I do not think AI will reach a point where you can make a tradeoff on things like pricing, settling a claim. That is still many years away."

The Five-Year Technology Rebuild

Abel's "narrow AI" philosophy is the top layer of a more fundamental infrastructure project at GEICO. The technology rebuild, now in approximately its fifth year, is restructuring GEICO's entire systems architecture from the ground up.

The legacy problem. GEICO historically operated on more than 600 legacy systems that, in Jain's words, "don't really talk to each other." Abel described the core issue directly: "We didn't have that ability to then use the information, get to the data." The rebuild aims to consolidate those 600-plus systems into 15 to 16 integrated platforms, a reduction of roughly 97% in application count.

From buyer to builder. The philosophical shift Abel articulated is from purchasing packaged technology solutions to building custom systems internally. "We recognized we were going to become a builder of technology rather than just a buyer of technology," Abel said. This is a meaningful commitment for a company that historically operated with minimal technology headcount relative to premiums written. GEICO opened a Palo Alto technology hub in October 2025, focused on AI/ML engineering, product development, platform engineering, and cybersecurity. GEICO's CTPO Vijay Raghavendra framed the ambition: "We're not just adopting new technologies; we are fundamentally reshaping how insurance is delivered in an AI-enabled world."

Cloud repatriation. One of the more counterintuitive elements of GEICO's tech strategy is the move away from public cloud. After migrating 600-plus applications to the cloud starting in 2013, GEICO found that cloud bills increased 2.5 times over a decade while reliability declined. The company has deployed over 1,000 Open Compute Project servers across two new colocation facilities and plans to scale to 3,000 servers, with a target of repatriating 50% of cloud workloads by 2028. GEICO reports a 50% reduction in compute cost per core and a 60-plus percent reduction in storage cost per gigabyte versus cloud and OEM pricing.

Workforce transformation. GEICO's headcount stood at 29,541 employees at year-end 2025, down from a peak of 42,156 in 2020. That is a reduction of nearly one-third, with 2023 alone accounting for roughly 7,700 departures (a 20% decline in a single year). The hiring is now shifting toward engineers and developers. Abel described a retraining effort that transforms operations staff into technology workers, consistent with the buyer-to-builder philosophy.

The Financial Mechanics of Funding a Tech Rebuild From Underwriting Profit

GEICO's full-year 2025 underwriting profit was $6.824 billion on $45.2 billion in premiums written. Even the Q1 2026 run rate of $1.4 billion per quarter, if annualized to roughly $5.6 billion, generates more than enough cash to fund technology investment in the hundreds of millions per year without touching the float or requiring capital allocation from Omaha. The beauty of the Berkshire structure is that GEICO's technology budget does not compete with other subsidiaries for capital; it comes directly from underwriting cash flow. This is the inverse of the typical startup carrier model where technology investment precedes underwriting profit. GEICO has the profit and is retrofitting the technology. The question is whether the retrofit can produce results before the competitive gap with Progressive and other telematics-first carriers widens further.

How Abel's Approach Contrasts With Peer Carrier AI Strategies

The May 2026 earnings season produced a clear spectrum of carrier AI philosophies. Abel's "narrow AI" stance sits at one end. At the other end are carriers deploying agentic systems at scale.

Travelers: Enterprise AI at scale. Travelers committed $1.5 billion in annual technology spend and deployed 10,000 Claude AI assistants through an Anthropic partnership. The deployment spans underwriting, claims, and operations. CEO Alan Schnitzer has framed AI as a competitive moat: the carriers that invest earliest and widest will compound advantages that laggards cannot close. This is the polar opposite of Abel's incremental, value-proven-first philosophy.

Chubb: Automation targets with headcount reduction. Greenberg set an explicit target of 85% back-office automation with nine AI projects in deployment and a stated goal of 1.5 combined ratio points in expense savings. Chubb's approach is top-down, CEO-directed, with quantified financial targets attached to each initiative. The Q1 2026 earnings call included an update on the claims AI mandate under newly appointed Global Claims Officer Kevin Rampe, with centralized deployment across 54 countries.

AIG: Agentic multi-model orchestration. AIG under CEO Peter Zaffino has been the most aggressive in deploying what the industry now calls "agentic AI," with multi-agent systems that orchestrate across underwriting, pricing, and risk selection. AIG's approach uses LLM agents to process submissions, extract data from unstructured documents, and make preliminary underwriting recommendations. The Q1 2026 earnings included a disclosure that the system had improved time-to-quote by 55% in tested lines.

Progressive: Data flywheel, not AI branding. Progressive rarely uses "AI" in its earnings communications. Instead, the competitive advantage is structural: 21 million telematics-connected policyholders feed a pricing model that continuously improves with each mile driven. The Q1 2026 result of 86.4% combined ratio with 9% PIF growth is the output of a decade-long data investment, not a recent AI deployment. Progressive proves that the most effective use of data in personal auto may not require the "AI" label at all.

Carrier	AI Philosophy	Q1 2026 CR	Key Metric
Berkshire/GEICO	Narrow AI, human governance	87.3%	1.5% premium growth
Progressive	Data flywheel (telematics)	86.4%	9% PIF growth
Chubb	85% automation target	84.0%	1.5pt CR savings goal
AIG	Agentic multi-model	87.3%	55% time-to-quote improvement
Travelers	Enterprise AI at scale	88.6%	10,000 Claude assistants

Insurance Float and the Investment Discipline Question

Berkshire's insurance float stood at $176.9 billion as of March 31, 2026, up from $176.4 billion at year-end 2025 and $171 billion at year-end 2024. The float represents premiums collected but not yet paid out in claims, and it forms the investable base that has been the economic engine of Berkshire's insurance model since Buffett acquired GEICO in 1996.

Under Abel's first quarter as CEO, the investment posture was conspicuously conservative. Berkshire held a record $397.4 billion in cash and short-term Treasury bills: $51.5 billion in cash and cash equivalents plus $339.3 billion in short-term U.S. Treasuries. The company was a net seller of $8.1 billion in equities during the quarter. Insurance investment income declined 7% to $2.68 billion, reflecting the conservative allocation.

For the insurance operations, the float economics remain attractive. Even at current short-term Treasury yields, $176.9 billion in float generating investment income of $2.68 billion per quarter (roughly $10.7 billion annualized) adds to the $1.72 billion in underwriting profit for a total insurance contribution approaching $4.4 billion per quarter. That is the economic power of a carrier that generates both underwriting profit and investment income on its float; most personal auto carriers generate one or the other, rarely both at this scale.

The investment discipline question under Abel is whether the conservative posture reflects a temporary holding pattern during the CEO transition or a permanent philosophical shift. Abel authorized $234 million in stock buybacks in Q1 2026, the first since May 2024, which suggests he is at least willing to deploy capital selectively. But the overwhelming signal is patience: $397 billion in cash earning Treasury yields while waiting for opportunities that meet Berkshire's return thresholds.

The Deepfake Demonstration and Cyber Risk Framing

One of the more unusual moments at the May 2026 annual meeting was Abel's live demonstration of a deepfake video that convincingly replicated Warren Buffett's voice and appearance using publicly available information. "That was done with zero input from Warren," Abel told the audience. "Voice, photo; we were able to obtain that with information that's out there, and replicate those actions and that voice."

Abel used the demonstration to frame cyber risk as a board-level concern: "That's what we're dealing with when we think of Berkshire and how we have to protect it every day. It can go to deepfakes, and they're using it to try to penetrate our business. It can be cyber attacks." This language positions cyber not as an underwriting opportunity but as an operational risk to be managed.

Jain's commentary on cyber insurance as a product line was notably cautious: "We have been slow in terms of entering that class of business as an underwriter. I find it very difficult to have some meaningful method to assess and model the aggregation." He added: "I'm not sure when, but I'm pretty certain that the day will come when we will have a fairly significant role to play in cyber." This patience mirrors the "narrow AI" philosophy: Berkshire will enter a market when it can model the risk adequately, not when the market is fashionable.

The Reinsurance Market Softening Signal

Abel used the annual meeting to deliver an unusually direct warning about reinsurance market conditions. "The reality is that as our insurance business softens, we cannot realize the value we should for the related risk," he said. Berkshire "will be much more cautious, specifically across the primary and reinsurance businesses."

The Q1 2026 data supports the concern. BH Reinsurance Group written premiums declined 2.3% to $5.992 billion, with RSUI premiums down 14% and NICO Primary premiums down 7%. The underwriting profit improvement came entirely from lower losses (the absence of California wildfire charges), not from pricing strength. When the largest reinsurer in the world signals caution about pricing adequacy, it is worth noting. This aligns with the broader soft market dynamics visible in the April 2026 renewal data, where Guy Carpenter's U.S. property cat ROL index fell 14%.

For reserving actuaries, the implication is that Berkshire's willingness to shrink reinsurance volume rather than chase inadequate rates provides a discipline anchor that smaller reinsurers may not follow. The gap between Berkshire's restraint and mid-market reinsurer behavior during softening phases has historically been a leading indicator of where reserve adequacy problems emerge two to three years later.

What This Means for Actuaries

GEICO's Q1 2026 results and Abel's technology strategy have several direct implications for actuarial practice.

Rate adequacy during technology transitions. GEICO's deliberate choice to suppress growth while rebuilding infrastructure creates a window where the carrier's rate adequacy may be stronger than its market position suggests. For state regulators reviewing GEICO's filings, the 87.3% combined ratio argues against rate increase approvals, but the underlying frequency and severity trends (both worsening) argue for them. The disconnect between profitability and trend is the same actuarial puzzle that Progressive's filings presented in 2018 to 2019 before the telematics advantage fully monetized.

Model validation under "narrow AI." Abel's temporal consistency test ("ask it to ignore today's information and just focus on yesterday; if we get the same answer, then we've safeguarded the system") is a practical, if simplified, form of model validation. For actuaries building AI governance frameworks, this suggests that Berkshire will apply a reproducibility standard to any AI system influencing underwriting decisions. That standard aligns with the stability and reproducibility requirements in the AI governance frameworks that SOA and CAS are developing, and it provides a data point for how the industry's largest carriers are interpreting ASOP No. 56 obligations.

Expense ratio monitoring for tech-investing carriers. GEICO's expense ratio jump from approximately 10.8% to 13.4% is partly advertising spend and partly technology investment. As more carriers enter multi-year technology transformations, the expense ratio becomes a less reliable comparability metric. Progressive's expense ratio in the 20% to 21% range includes agent commissions that GEICO does not pay. Chubb's expense ratio includes the overhead of 54-country operations. Comparing expense ratios across carriers undergoing different technology transformations requires normalizing for the investment component, a task that current statutory filings make difficult because technology spending is not separately disclosed.

Reserve implications of frequency reversal. The reversal of favorable claim frequency trends at GEICO, with frequency and severity both increasing, has direct reserve implications. Prior-year reserve estimates that assumed continued frequency favorability will need re-evaluation. For actuaries at competitor carriers watching the GEICO disclosure, the frequency reversal is an industry signal, not a GEICO-specific phenomenon. If the post-pandemic frequency tailwind is ending across personal auto, it affects every carrier's loss picks for accident years 2024 through 2026.